Assignment1AnswerStefanLeachu.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# IDS Assignment 1\n",
    "Document your results as well as the way you obtained them in this jupyter notebook. A seperate report (pdf, word etc.) is _not_ required. However, it is necessary that you provide the python code leading to your results as well as textual answers to the assignment questions in this notebook. \n",
    "\n",
    "Do not change the general structure of this notebook, but you can add further markdown or code cells to explain your solutions if necessary. In the end, submit this file in moodle."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Get to know the data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 1 (Indicating type): \n",
    "1. Identify precise type of each attribute. Note that the right answers are like: “numerical-continuous-ratio”, just Numerical or Categorical is not enough."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation:\n",
    "\n",
    "The types of the attributes of our dataset are:\n",
    "- <code>age</code> : numerical continuous ratio - can not be counted; has true 0 and can be * or /\n",
    "- <code>workclass</code> : categorical nominal - has no order nnnnnafasdfasdvakfnsadfpnö\n",
    "- <code>education</code> : categorical ordinal - can be ordered obviously\n",
    "- <code>education-num</code> : categorical ordinal - e.g. two bachelor degrees can not be added up, that why it is not numeric; the numbers are only another synonymous label for the categorical *education*; theoretically same attribute as *education*, therefore no new information/ no information gain which means that it is enough to consider always only one of these two attributes \n",
    "- <code>marital-status</code> : categorical nominal - ordering would be subjective; no status is 'better' than another\n",
    "- <code>occupation</code> : categorical nominal - ordering would be subjective; no occupation is 'better' than another\n",
    "- <code>relationship</code> : categorical nominal - ordering would be subjective; no status is 'better' than another\n",
    "- <code>race</code> : categorical nominal - no race should be better than another\n",
    "- <code>sex</code> : categorical nominal - no sex should be better than another\n",
    "- <code>hours-per-week</code> : numerical continuous ratio - can not be counted; has true 0 and can be * or /\n",
    "- <code>native-country</code> : categorical nominal - no native-country should be better another\n",
    "- <code>label</code> : categorical nominal - label says already that its only a label like red or green; is not ordinal because these are two categories without judging"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 2 (Removing outliers):\n",
    "2. (a) Explore into \"age\" and \"hours-per-week\" and identify outliers (Boxplot returns whiskers). After identifying outliers, remove them (do it just once on the data set). Now you should have two data sets (cleaned and original)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAEICAYAAABRSj9aAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAF2lJREFUeJzt3X2UXPV93/H3V9Jq14CxeBALCJCcoGCxPnXibAwuEAgiBQIFnR5ckAGDj1yKYoTt1I1d6ySmOYfE6REhQXVwlQqbludQ2ygcQ8ODIVYb7ArbdSI2BtVY6AGWBSTsyFlJSN/+MVd4WGZH0txdXe3V+3XOnJ37+PvO7O5nfvObO/dGZiJJqq9JVRcgSRpfBr0k1ZxBL0k1Z9BLUs0Z9JJUcwa9JNWcQS/thYg4KyLWV11HOxFxdUSsrLqOkSLixxFxTtV1HIgM+ppp9c+0v/7ja3QRcUNE3FF1HaoHg14di4gpB2Lb0kRj0B+AImJORDwREZsjYnVEXNS07ImI+FjT9FveDURERsTHI+I54LlouDkiXo6I1yPiBxHx3lHaPSsi1kfE5yLileLdx+VNy7sjYklEvBARgxHxpYh4x4htPxMRLwFfbrH/j0bEXzVNr4mI+5qm10XELxf33xMRj0TEaxHxw4j413tSR4s2r4+IZyLiuBbLro6IlcW+NkXE8xFxftPyYyNiRVHDmoj4N8X884DPAZdGxD9GxP8dpe3jI+KrETEUEa9GxH8eZb12j/WCiPheRPykeH5uaFo2q/h9X1U8F69ExOKm5ZMi4rMR8f+K9u+LiMObll8ZEWuLZW9up33PoD/AREQX8FfAXwNHAYuAOyPipL3YzTzgFOBk4F8Avw78EjANuBR4tc22RwNHAjOAq4BlTW3/cbGfXwZOLNb5/RHbHg7MBK5pse8ngTOKADoG6AJOKx73LwCHAD+IiIOBR4C7aDwH84E/j4i+PayDYp+/B1wNnJmZo43bnwL8sHjM/wlYHhFRLLsbWA8cC1wC/GFEzM3Mh4E/BO7NzEMy830t2p4MPAisBWYVNd7TYr3dPdYtwEdo/O4uABZGxLwRuzkdOAmYC/x+RMwp5l9P42/hzOIxbAK+WLR7MnArcGWx7AjgbS+G2kcy01uNbsCPgX8ENjfdfgasLJafAbwETGra5m7ghuL+E8DHmpZdvWvbYjqBs5umzwaeBU5t3ucotZ0FvAEc3DTvPuD3gKAROr/YtOyDwPNN224DenbTxjrg/cBlwDLgO8B7gI8CK4p1LgW+NWK7/wJ8fg/r2AD8CbASeFebWq4G1jRNH1Q8f0cDxwM7gHc2Lf8j4CvF/RuAO9rs+4PAEDBllHZX7u6xjrLfPwVuLu7PKuo9rmn5d4DLivsDwNymZccA24EpNF4Y72ladnDx+zun6v+RA/HmOGc9zcvMR3dNRMTVwK7hmGOBdZm5s2n9tTR6hHtq3a47mfl4MWTwReCEiPga8GkaPcRnmtY7pLi7KTO3jGj7WGA6jSB8+ucdXgKY3LTuUGYONz2uh2i8cAH828y8k0av/iwaPfEnabzQnUkjGJ8s1p0JnBIRm5v2PQX473tYxzQa7yguzczXWz5DP/dS03Pws2Kfh9Do4b6WmT8d8Vz072Z/uxwPrM3MN3azXrvHSkScAnwBeC8wFegG/nK0x0Cj07DrdzkT+FpENP8t7QB6Kf7Ods3MzC0R0e6dnsaRQzcHno3A8RHR/Ls/gUYvFRq92YOalh3dYh9vOeVpZt6Smb8K9NEY8vj3mflCNoYdDmkKeYDDiuGE5rY3Aq8A/wT0Zea04vauEduObPf8pjbuLGbvCvozivtP0gj6M/l50K8DnmxqZ1qxj4V7WMcm4ELgyxFxWovnZ09sBA6PiHeOeC52/R52d1rZdTReWHfXWWv3WKExpLMCOD4z3wV8icYL255YB5w/Yt89mbkBeJHGixEAEXEQjRc3VcCgP/B8m0aY/25EdEXEWcC/5Ofju98H/lVEHBQRJwIL2u0sIn4tIk4pxv63AMM0enXt/MeImBoRZ9AIzL8s3mH8BXBzRBxV7HtGRJy7l4/vSeA3gHdkY9z8W8B5NELme8U6DwK/VHxY2FXcfi0i5uxpHZn5BHA5jR7tKXtZI5m5DvjfwB9FRE9E/DMaz/WuF6xBYNaIF+Rm36ERpl+IiIOLfbR60Rn1sRbL30njncVwRHwA+PBePIwvATdGxEyAiJgeERcXy+4HLoyI0yNiKvAHmDeV8Yk/wGTmNuAi4Hwavdc/Bz6Smf9QrHIzjbHUQeB2fh48ozmURjBuojH08CqwpM36LxXrbiz2fW1T258B1gBPRcRPgEdpfAi4N4/vWRqfUXyrmP4J8CPgf2XmjmLeT2l8iHxZUcdLND6A7d6bOjLzEYqx/4j41b2pszCfxjj4RuBrNMbNHymW7Ro+eTUivtui7R00XqBPBF6g8aHupS3W291j/W3gDyLipzTG1e8buY82/ozGu4G/LrZ/isaHz2TmauDjNN4xvEjjd75ff9GsziLTC49o3yjePdyRmR59Ie1D9uglqeYMekmqOYduJKnm7NFLUs3tF1+YOvLII3PWrFlVlyFJE8rTTz/9SmZO3916+0XQz5o1i1WrVlVdhiRNKBGxdk/Wc+hGkmrOoJekmjPoJanmdhv0EXFbNC4q8fdN8w4vLmTwXPHzsGJ+RMQt0biIwg8i4v3jWbwkaff2pEf/FRonhWr2WeCxzJwNPFZMQ+P8KbOL2zU0LjwgSarQboM+M/8GeG3E7ItpnPCK4ue8pvn/LRueAqYVV/qRJpSIeNtNmqg6HaPvzcwXAYqfRxXzZ9B0sQEaZ6treUGLiLgmIlZFxKqhoaEOy5DGXnOoX3zxxS3nSxPJWH8Y2+o/oeU5FjJzWWb2Z2b/9Om7Pd5fqsQDDzxQdQlSaZ0G/eCuIZni58vF/PU0XVWGxsWAN3ZenlSdvr4+1q5dS19f3+5XlvZjnQb9CuCq4v5VwANN8z9SHH1zKvD6riEeaaJZs2YNM2fOZM2aNVWXIpWyJ4dX3g38LXBSRKyPiAU0Lib8mxHxHPCbxTTAN2hczWcNjasO/fa4VC3tA1u3buWCCy5g69atVZcilbJfnKa4v78/PdeN9hftPnTdH/5fpF0i4unM7N/den4zVhpFb28vAwMD9Pb2Vl2KVMp+cfZKaX8zefJkBgcHmTNnzpvTO3bsqLgqqTP26KUWduzYwcKFC9m8eTMLFy405DWhOUYvjTBp0iRmzJjBhg0byEwi4s3pnTt3Vl2e9CbH6KUOZSbr16/n2muvZfPmzVx77bWsX7/eD2I1YRn00gjd3d2cdtpp3HbbbUybNo3bbruN0047je7u7qpLkzpi0EsjbNu2jY0bN/LQQw+xbds2HnroITZu3Mi2bduqLk3qiEfdSCOcfPLJzJs3j0WLFjEwMMCcOXP48Ic/zNe//vWqS5M6Yo9eGmHx4sXcddddLF26lOHhYZYuXcpdd93F4sWLqy5N6ohBL40wf/58Zs+ezdy5c5k6dSpz585l9uzZzJ8/v+rSpI4Y9NIIixYt4vHHH2fJkiVs2bKFJUuW8Pjjj7No0aKqS5M64nH00gg9PT0tT2TW3d3N8PBwBRVJrXkcvdSh5pC///77W86XJhKDXhpFd3c3l1xyicfPa8Iz6KVRnHvuuQwNDXHuuedWXYpUisfRS6NYsWIFXs9YdWCPXmrjjjvuqLoEqTSDXmrjiiuuqLoEqTSDXmphypQp3HTTTWzZsoWbbrqJKVMc5dTE5XH00gg9PT3MnDmT55577s3z0c+ePZu1a9d6HL32Kx5HL3XozDPP5Nlnn33L+eifffZZzjzzzKpLkzri+1FphA0bNtDV1cWtt97KrbfeCkBXVxcbNmyouDKpM/bopRFWr17N9u3b6evrY+3atfT19bF9+3ZWr15ddWlSR+zRS6NYvXo1M2fOrLoMqTR79FIbt9xyS9UlSKUZ9NIo+vr6+OQnP0lfX1/VpUilGPRSG88//3zVJUilOUYvjcIxetWFPXpJqjmDXhrhrrvu2qv50v7OoJdGuPHGG1m8eDF9fX1MmjSJvr4+Fi9ezI033lh1aVJHSo3RR8SngI8BCfwd8FHgGOAe4HDgu8CVmbmtZJ3SPvPMM8/ws5/9jOXLl3P66aezcuVKFixYwI9//OOqS5M60nHQR8QM4Hrg5Mz8p4i4D7gM+C3g5sy8JyK+BCwAbh2TaqV9YOrUqTz//POcffbZb5nvJQU1UZUdupkCvCMipgAHAS8CZwO7rqh8OzCvZBvSPtV8EfCjjz665XxpIuk46DNzA7AEeIFGwL8OPA1szsw3itXWAzPKFilVoa+vj5dfftkvTGnC6zjoI+Iw4GLg3cCxwMHA+S1WbXnC+4i4JiJWRcSqoaGhTsuQxsXkyZNZunQpw8PDLF26lMmTJ1ddktSxMh/GngM8n5lDABHxVeCfA9MiYkrRqz8O2Nhq48xcBiyDxoVHStQhjbkdO3a8bYxemqjKjNG/AJwaEQdFRABzgWeAbwKXFOtcBTxQrkSpOkcccUTVJUillRmj/zaND12/S+PQykk0euifAX4nItYARwDLx6BOqRKvvvpq1SVIpZU66iYzP5+Z78nM92bmlZm5NTN/lJkfyMwTM/NDmemhCpqQFi5cyObNm1m4cGHVpUileHFwaYSIoKuri0mTJrF161a6u7vZuXMn27dvZ3/4f5F28eLgUgnbt29n2rRpDAwMMG3aNLZv3151SVLHPE2xNIrBwUHmzJlTdRlSafbopTauv/76qkuQSjPopTa8ZqzqwKCXRuFRN6oLj7qRRmh8/6/xMzPf/Al41I32Kx51I5WUmXz605823DXhGfTSCM3nnV+yZEnL+dJEYtBLIzSfd/7yyy9vOV+aSAx6aRR9fX3cfffdno9eE55BL7XQ3d39lvPRO2yjicxvxkotbN261fPRqzbs0UtSzRn0klRzDt1Io2g+fn7Xl6ikicigl0ZhuKsuHLqRpJoz6CWp5gx6aRQXXXQRQ0NDXHTRRVWXIpXiGL00ihUrVjB9+vSqy5BKs0cvjSIiePjhh/1QVhOePXppFJnJeeedV3UZUmn26CWp5gx6Sao5g14axeTJk3niiSeYPHly1aVIpThGL41ix44dnHXWWVWXIZVmj16Sas6gl6SaM+ilNu65556qS5BKM+ilNi677LKqS5BKM+ilUfT29jIwMEBvb2/VpUillAr6iJgWEfdHxD9ExEBEfDAiDo+IRyLiueLnYWNVrLQvDQ4OMmfOHAYHB6suRSqlbI/+z4CHM/M9wPuAAeCzwGOZORt4rJiWJpyenh6eeuopenp6qi5FKqXjoI+IQ4FfB5YDZOa2zNwMXAzcXqx2OzCvbJHSvtTd3Q3A8PAwp556KsPDw2+ZL000Zb4w9QvAEPDliHgf8DTwCaA3M18EyMwXI+KoVhtHxDXANQAnnHBCiTKkPVfmTJRbt27d4+2brzcrVa3M0M0U4P3ArZn5K8AW9mKYJjOXZWZ/ZvZ7zm/tK5m5R7frrrvuzR58d3c311133R5va8hrfxOd/lFGxNHAU5k5q5g+g0bQnwicVfTmjwGeyMyT2u2rv78/V61a1VEd0niKCINb+62IeDoz+3e3Xsc9+sx8CVgXEbtCfC7wDLACuKqYdxXwQKdtSJLKK3tSs0XAnRExFfgR8FEaLx73RcQC4AXgQyXbkCSVUCroM/P7QKu3DXPL7FeSNHb8Zqwk1ZxBL0k1Z9BLUs0Z9JJUcwa9JNWcQS9JNWfQS1LNGfSSVHMGvSTVnEEvSTVn0EtSzRn0klRzBr0k1ZxBL0k1Z9BLUs0Z9JJUcwa9JNWcQS9JNVf2mrFSZQ4//HA2bdo07u1ExLi3cdhhh/Haa6+Nezs6MBn0mrA2bdpEZlZdxpjYFy8mOnA5dCNJNWfQS1LNGfSSVHMGvSTVnEEvSTVn0EtSzRn0klRzBr0k1ZxBL0k1Z9BLUs0Z9JJUc6WDPiImR8T3IuLBYvrdEfHtiHguIu6NiKnly5QkdWosevSfAAaapv8YuDkzZwObgAVj0IYkqUOlgj4ijgMuAP5rMR3A2cD9xSq3A/PKtCFJKqfsaYr/FPhd4J3F9BHA5sx8o5heD8xotWFEXANcA3DCCSeULEMHovz8oXDDu6ouY0zk5w+tugTVWMdBHxEXAi9n5tMRcdau2S1WbXnC8MxcBiwD6O/vr8dJxbVv3fD6uDcREbU5570OXGV69KcBF0XEbwE9wKE0evjTImJK0as/DthYvkxJUqc6HqPPzP+Qmcdl5izgMuDxzLwc+CZwSbHaVcADpauUJHVsPI6j/wzwOxGxhsaY/fJxaEOStIfG5JqxmfkE8ERx/0fAB8Ziv5Kk8vxmrCTVnEEvSTVn0EtSzRn0klRzBr0k1ZxBL0k1Z9BLUs0Z9JJUcwa9JNWcQS9JNWfQS1LNGfSSVHMGvSTVnEEvSTVn0EtSzRn0klRzBr0k1dyYXGFKqpuIeNv9zKyqHKkUg14HlOYAH89tfVHQ/sSg1wFlTwK4XaAb4JqIHKOXRtHV1cXKlSvp6uqquhSpFHv00ii2bdv25s8yQz5S1Qx6aRSGu+rCoRtJqjmDXpJqzqCXpJoz6CWp5gx6Sao5g16Sas6gl6SaM+glqeY6DvqIOD4ivhkRAxGxOiI+Ucw/PCIeiYjnip+HjV250r51xx13VF2CVFqZHv0bwL/LzDnAqcDHI+Jk4LPAY5k5G3ismJYmpCuuuKLqEqTSOg76zHwxM79b3P8pMADMAC4Gbi9Wux2YV7ZISVLnxmSMPiJmAb8CfBvozcwXofFiABw1Fm1IkjpTOugj4hDgfwCfzMyf7MV210TEqohYNTQ0VLYMacz19PTw1FNP0dPTU3UpUimlgj4iumiE/J2Z+dVi9mBEHFMsPwZ4udW2mbksM/szs3/69OllypDGxfDwMKeeeirDw8NVlyKVUuaomwCWAwOZ+SdNi1YAVxX3rwIe6Lw8SVJZZc5HfxpwJfB3EfH9Yt7ngC8A90XEAuAF4EPlSpQkldFx0GfmSmC0KzPM7XS/0v7k/vvv55JLLqm6DKkUvxkrtWHIqw4MekmqOYNekmrOoJekmjPoJanmDHqpjYMPPrjqEqTSDHqpjS1btlRdglSaQS9JNWfQS1LNGfRSG729vVWXIJVm0EttDA4OVl2CVJpBL7Xx4IMPVl2CVJpBL7Vx4YUXVl2CVJpBL0k1Z9BLUs0Z9NIourq6WLlyJV1dXVWXIpVS5gpTUq1t376d008/veoypNLs0UtSzRn0klRzBr3UhsfRqw4MeqkNj6NXHRj0UhvHH3981SVIpRn0Uhvr1q2rugSpNINeamPBggVVlyCVZtBLbSxfvrzqEqTSDHqpjfnz51ddglSaQS+1cffdd1ddglSaQS+1cemll1ZdglSaQS+1ce+991ZdglSaQS+18alPfarqEqTSDHqpjZtvvrnqEqTSDHqpjVtuuaXqEqTSxiXoI+K8iPhhRKyJiM+ORxvSvnD99ddXXYJU2pgHfURMBr4InA+cDMyPiJPHuh1pvE2aNIlHH32USZN846uJbTz+gj8ArMnMH2XmNuAe4OJxaEcaVzt37uScc85h586dVZcilTIeQT8DaD4T1Ppi3ltExDURsSoiVg0NDY1DGVJnuru7Aejt7WVgYIDe3t63zJcmmvEI+mgxL982I3NZZvZnZv/06dPHoQypM8PDw3R3dzM4OMicOXMYHByku7ub4eHhqkuTOjIeFwdfDzSfxPs4YOM4tCONG0NddTIePfr/A8yOiHdHxFTgMmDFOLQjSdoDY96jz8w3IuI64H8Ck4HbMnP1WLcjSdoz4zF0Q2Z+A/jGeOxbkrR3PEBYkmrOoJekmovMtx35uO+LiBgC1lZdh9TCkcArVRchjWJmZu72+PT9Iuil/VVErMrM/qrrkMpw6EaSas6gl6SaM+il9pZVXYBUlmP0klRz9uglqeYMekmqOYNeaiEibouIlyPi76uuRSrLoJda+wpwXtVFSGPBoJdayMy/AV6rug5pLBj0klRzBr0k1ZxBL0k1Z9BLUs0Z9FILEXE38LfASRGxPiIWVF2T1ClPgSBJNWePXpJqzqCXpJoz6CWp5gx6Sao5g16Sas6gl6SaM+glqeb+P8MtGK+epQBVAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# your code\n",
    "\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "#read data from csv\n",
    "df = pd.read_csv('adult.data-3.csv')\n",
    "\n",
    "#plot age-boxplot\n",
    "bpAge = plt.boxplot(df['age'])\n",
    "plt.title(\"Age not cleaned\")\n",
    "plt.figure()\n",
    "\n",
    "#plot hours-per-week-boxplot\n",
    "bpHours = plt.boxplot(df['hours-per-week'])\n",
    "plt.title(\"Hours-per-week not cleaned\")\n",
    "plt.figure()\n",
    "\n",
    "dfClean = df\n",
    "\n",
    "#clean age\n",
    "##remove lower outliers\n",
    "dfClean = dfClean.drop(dfClean[dfClean['age']<bpAge['whiskers'][0].get_ydata()[-1]].index)\n",
    "##remove upper outliers\n",
    "dfClean = dfClean.drop(dfClean[dfClean['age']>bpAge['whiskers'][1].get_ydata()[-1]].index)\n",
    "\n",
    "#clean hours-per-week\n",
    "##remove lower outliers\n",
    "dfClean = dfClean.drop(dfClean[dfClean['hours-per-week']<bpHours['whiskers'][0].get_ydata()[-1]].index)\n",
    "##remove upper outliers\n",
    "dfClean = dfClean.drop(dfClean[dfClean['hours-per-week']>bpHours['whiskers'][1].get_ydata()[-1]].index)\n",
    "\n",
    "\n",
    "#for visual comparison boxplots of cleaned data\n",
    "##plot age-boxplot cleaned\n",
    "plt.boxplot(dfClean['age'])\n",
    "plt.title(\"Age cleaned\")\n",
    "plt.figure()\n",
    "##plot hours-per-week-boxplot cleaned\n",
    "plt.boxplot(dfClean['hours-per-week'])\n",
    "plt.title(\"Hours-per-week cleaned\")\n",
    "plt.figure()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "2. (b) Draw Boxplot of the cleaned data set for \"hours-per-week\". Is there still any outlier? If so, explain why?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# your code\n",
    "\n",
    "#plot hours-per-week-boxplot cleaned\n",
    "plt.boxplot(dfClean['hours-per-week'])\n",
    "plt.title(\"Hours-per-week cleaned\")\n",
    "plt.figure()\n",
    "\n",
    "### only to demonstrate the outlier fact; see explanation\n",
    "#remove \"lower\" outliers\n",
    "dfWrong = dfClean.drop(dfClean[dfClean['hours-per-week']<=bpHours['whiskers'][0].get_ydata()[-1]].index)\n",
    "#remove \"upper\" outliers\n",
    "dfWrong = dfClean.drop(dfClean[dfClean['hours-per-week']>=bpHours['whiskers'][1].get_ydata()[-1]].index)\n",
    "plt.boxplot(dfWrong['hours-per-week'])\n",
    "plt.title(\"Hours-per-week cleaned wrongly\")\n",
    "plt.figure()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: No, there are not any outliers. Caused by the fact that a boxplot is a relative measure (the boxplot depends on values like median, IQR, quartiles), it is possible that there are not outliers like in our case. Especially this means that there are not any outliers in the data set (hours-per-week cleaned) beyond the 1.5*IQR mark ((upper and lower whiskers) to become, so called, fliers.\n",
    "\n",
    "If the data set is modified incorrectly (demonstrated by the set *dfWrong*) which means that the lower and upper whiskers are deleted wrongly, then we can see outliers again. Now, this is caused by different the destribution of the data in *dfWrong* in comparison to the data set *dfClean*."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 3 (Basic visualization):\n",
    "3. (a) Visualize mean and median of \"age\", and \"hours-per-week\" per \"sex\" by separate plots for the cleaned and the original data sets (there should be 8 plots, 4 plots for the original data set and 4 plots for the cleaned data set). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean   male   age original: 41.12117637807659\n",
      "Mean   male   age cleaned : 40.9983105104037\n",
      "Median male   age original: 40.0\n",
      "Median male   age cleaned : 40.0\n",
      "Mean   female age original: 37.57591725214676\n",
      "Mean   female age cleaned : 37.77269372693727\n",
      "Median female age original: 36.0\n",
      "Median female age cleaned : 37.0\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.JointGrid at 0x1a19c43400>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# your code\n",
    "# Necessary: Please run Q2 before, for obtaining the cleaned data\n",
    "\n",
    "import seaborn as sns \n",
    "import numpy as np\n",
    "\n",
    "#plot mean-age-original-barplot\n",
    "sns.barplot(x ='sex', y='age', data=df, estimator=np.mean)\n",
    "plt.title(\"Mean age - Original Data\")\n",
    "plt.figure()\n",
    "\n",
    "#plot mean-age-clean-barplot\n",
    "sns.barplot(x ='sex', y='age', data=dfClean, estimator=np.mean)\n",
    "plt.title(\"Mean age - Clean Data\")\n",
    "plt.figure()\n",
    "\n",
    "#plot median-age-original-barplot\n",
    "sns.barplot(x='sex', y='age', data=df, estimator=np.median)\n",
    "plt.title(\"Median age - Original Data\")\n",
    "plt.figure()\n",
    "\n",
    "#plot median-age-clean-barplot\n",
    "sns.barplot(x='sex', y='age', data=dfClean, estimator=np.median)\n",
    "plt.title(\"Median age - Clean Data\")\n",
    "plt.figure()\n",
    "\n",
    "#plot mean-hours-per-week-original-barplot\n",
    "sns.barplot(x ='sex', y='hours-per-week', data=df, estimator=np.mean)\n",
    "plt.title(\"Mean hours-per-week - Original Data\")\n",
    "plt.figure()\n",
    "\n",
    "#plot mean-hours-per-week-cleam-barplot\n",
    "sns.barplot(x ='sex', y='hours-per-week', data=dfClean, estimator=np.mean)\n",
    "plt.title(\"Mean hours-per-week - Clean Data\")\n",
    "plt.figure()\n",
    "\n",
    "#plot median-hours-per-week-original-barplot\n",
    "sns.barplot(x='sex', y='hours-per-week', data=df, estimator=np.median)\n",
    "plt.title(\"Median hours-per-week - Original Data\")\n",
    "plt.figure()\n",
    "\n",
    "#plot median-hours-per-week-clean-barplot\n",
    "sns.barplot(x='sex', y='hours-per-week', data=dfClean, estimator=np.median)\n",
    "plt.title(\"Median hours-per-week - Clean Data\")\n",
    "plt.figure()\n",
    "\n",
    "#numerical values for 3b\n",
    "##drop females in original and clean data\n",
    "dfM = df.drop(df[df['sex']=='Female'].index)\n",
    "dfMC = dfClean.drop(dfClean[dfClean['sex']=='Female'].index)\n",
    "print(\"Mean   male   age original:\", np.mean(dfM['age']))\n",
    "print(\"Mean   male   age cleaned :\", np.mean(dfMC['age']))\n",
    "print(\"Median male   age original:\", np.median(dfM['age']))\n",
    "print(\"Median male   age cleaned :\", np.median(dfMC['age']))\n",
    "##drop males in original and clean data\n",
    "dfF = df.drop(df[df['sex']==\"Male\"].index)\n",
    "dfFC = dfClean.drop(dfClean[dfClean['sex']==\"Male\"].index)\n",
    "print(\"Mean   female age original:\", np.mean(dfF['age']))\n",
    "print(\"Mean   female age cleaned :\", np.mean(dfFC['age']))\n",
    "print(\"Median female age original:\", np.median(dfF['age']))\n",
    "print(\"Median female age cleaned :\", np.median(dfFC['age']))\n",
    "\n",
    "#visualization of the females orginal data\n",
    "sns.jointplot(x=['age'], y=['hours-per-week'], data=dfF, kind=\"kde\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "3. (b) Explain how mean and median of \"age\" per \"sex\" change when you remove the outliers."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: With the visualization in bar plots of the mean and median, we can compare proportinally *one* statistical feature splitted per sex. However, for comparing the mean and median now exactly, we have to look at the numerical values, because these can not be detected in the bar plots. Therefore, the values are printed.  \n",
    "With the numerical values you can see that only the median of the male people does not change, the rest changes. In case of *age* of the *males* the mean is a little bit lower. This means that there are \"a lot\" of *males* with a higher *age* and/ or with outlyling working *hours-per-week* in the not cleaned data set. By cleaning these, outlying values are deleted, so that the mean shrinks. The median has the same value, 40 years. In fact this means that there are before and after cleaning still 50% of *males* in the data set over and under 40 years old.  \n",
    "In contrast the mean and median of *age* of the *females* increased. This fact is caused by cleaning the data set by *age*, but in addation by *hours-per-week* too. In the age-boxplot from question 2 we can identify that there are only outliers in the feature of *age* above the upper whisker. Intuitivly, the mean and median should be decrease, because we drop only values above the mean and median. But due to the fact of cleaning the feature *hours-per-week* too, there is the possibility of increasing the mean and median. In fact these increased values means that there are \"a lot\" of young *females* working less than the lower whisker values of this feature. Maybe this is caused by the higher family wish (pregnancy) in this range of age. (e.g. visualizable wih jointplot - positive/ right skewed distribution at *age* seeable, and raised number of young females working about 20h per week)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 4 (Distribution):\n",
    "4. (a) Explore into distribution of \"age\" in the original data set. Does it have any well-known distribution (normal, uniform, skewed, ...)? If so, what are the main statistical features (mean, median, and mode) of this specific type of distribution."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean   age original data: 40.10296508043271\n",
      "Median age original data: 39.0\n",
      "Mode   age original data: ModeResult(mode=array([37]), count=array([532]))\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# your code\n",
    "# no code from above for execution needed --- new section\n",
    "\n",
    "import pandas as pd\n",
    "import seaborn as sns \n",
    "import numpy as np\n",
    "from scipy import stats\n",
    "\n",
    "#read data from csv\n",
    "df = pd.read_csv('adult.data-3.csv')\n",
    "\n",
    "sns.distplot(df['age'])\n",
    "\n",
    "print(\"Mean   age original data:\", np.mean(df['age']))\n",
    "print(\"Median age original data:\", np.median(df['age']))\n",
    "print(\"Mode   age original data:\", stats.mode(df['age']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: The distribution is a well-known positive/ right skewed distribution. This can be seen visual in the displot and, as in the most cases of right skewed distributions, in the relation of mean, median and mode. These three values occur in the sequence of mode<median<mean in right skewed distribution. Also the fact of median&#8800;mean indicates that it can not be a normal distribution."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "4. (b) Explore into distribution of \"age\" and \"hours-per-week\" together. Explain how the data is distributed with respect to these two attributes in the original data sets (using Jointplot)?\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.JointGrid at 0x1a1b917898>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# your code\n",
    "\n",
    "sns.jointplot(x=['age'], y=['hours-per-week'], data=df, kind=\"kde\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: We can see that most of the working people (20-65 years) work 40 hours-per-week (darkblue). A few of the ~20 years old people work only 20 hours-per-week (light blue). Maybe like 'HiWis' during their studies. In addition we can see that there is an higher amount of middle aged people (~40years) which are working more than 40 hours per week, maybe higher job positions which are characterized by this group of age."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Building models to perform predictions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 5 (Decision Tree):\n",
    "5. (a) Consider all the categorical attributes except label as descriptive features and label (income) as target feature. Make two decision trees (one based on Entropy and another based on Gini)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:25: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n"
     ]
    }
   ],
   "source": [
    "# your code\n",
    "# no code from above for execution needed --- new section\n",
    "\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn import tree\n",
    "from subprocess import check_output\n",
    "import seaborn as sns \n",
    "import numpy as np\n",
    "from DecisionTree import DecisionTree\n",
    "\n",
    "#read data from csv\n",
    "df = pd.read_csv('adult.data-3.csv')\n",
    "\n",
    "categorical_features = df[['workclass','education','marital-status','occupation','relationship','race','sex','native-country','label']]\n",
    "columns = categorical_features.columns\n",
    "\n",
    "#all columns except the last one are descriptive by default\n",
    "descriptive_features = columns[:-1]\n",
    "#last column is considered as label\n",
    "label = columns[-1]\n",
    "\n",
    "#converting all columns to string \n",
    "for column in columns:\n",
    "    categorical_features[column] = categorical_features[column].astype(str)\n",
    "   \n",
    "data_descriptive = categorical_features[descriptive_features].values\n",
    "data_label = categorical_features[label].values\n",
    "\n",
    "decisionTreeE = DecisionTree(data_descriptive.tolist(), descriptive_features.tolist(), data_label.tolist(), \"entropy\")\n",
    "decisionTreeE.id3(0,0)\n",
    "#decisionTreeE.print_visualTree()\n",
    "\n",
    "decisionTreeG = DecisionTree(data_descriptive.tolist(), descriptive_features.tolist(), data_label.tolist(), \"gini\")\n",
    "decisionTreeG.id3(0,0)\n",
    "#decisionTreeG.print_visualTree()\n",
    "\n",
    "#print(\"Entropy: \", format(decisionTreeE.entropy))\n",
    "#print(\"Gini   : \", format(decisionTreeG.gini))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "5. (b) What are the best attributes (based on Gini and Entropy) for splitting the trees in the third round of ID3?   "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: In both cases (entropy and gini) the best attribute for splitting the trees in the third round of ID3 is *occupation*. Beginning from the root, the first one is *relationship*, second one education and the third one *occupation*. Furthermore, both decision trees, based on entropy and gini, are the same."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "5. (c) Prune the tree which is made based on Entropy by 7000 as minimum number of samples. Identify which value of the first node/attribute is chosen to be split in the second round of ID3 and explain why?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# your code\n",
    "\n",
    "decisionTreeE.id3(0,7000)\n",
    "#decisionTreeE.print_visualTree()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: The chosen first node/ attribute for splitting in the secound round of ID3 has got the value *Husband*. This node is chosen because its entropy is the lowest, respectivly its information gain is the highest in this step of choosing in the alogrithm. Due to the fact that this subtree is the same as in the not pruned tree, the criteria of 7000 minimum samples is fulfilled in particular."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 6 (Regression):\n",
    "\n",
    "6. (a) Create two sets of independent variables (1) and (2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "# your code\n",
    "# no code from above for execution needed --- new section\n",
    "\n",
    "import pandas as pd\n",
    "\n",
    "#read data from csv\n",
    "df = pd.read_csv('adult.data-3.csv')\n",
    "\n",
    "#split original data into independent variables\n",
    "dataset1 = df[['age','marital-status','sex']]\n",
    "dataset2 = df[['workclass','education','hours-per-week']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "6. (b) Which attributes are suitable as an input for the logistic regression and which need to be modified first? Why? "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: \n",
    "\n",
    "Suitable as an input for the logistic regression are all continuous descriptive features. These are in our case: age, education-num, hours-per-week. And a binary categorical target feature, here label.  \n",
    "All remaning attributes (workclass, education, marital-status, occupation, relationship, race, sex, native-country)\n",
    "are categorical descriptive features. Therefore those have to be encoded, e.g. with one-hot-encoding, to become continuous descriptives features, in other words to become a valid input for the logistic resgression."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "# your modification\n",
    "\n",
    "#original data\n",
    "dataset = df[['age','workclass','education-num','marital-status','occupation','relationship','race','sex','hours-per-week','native-country']]\n",
    "dataset = pd.get_dummies(dataset, columns = ['workclass','marital-status','occupation','relationship','race','sex','native-country'])\n",
    "##esthetic - move target feature to last position [-1] \n",
    "label = df['label']\n",
    "dataset = pd.concat([dataset,label], axis=1)\n",
    "\n",
    "#dataset1 - encode columns\n",
    "dataset1 = pd.get_dummies(dataset1, columns=['marital-status','sex'])\n",
    "\n",
    "#dataset2 - encode columns\n",
    "dataset2 = pd.get_dummies(dataset2, columns=['workclass','education'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "6. (c) Train three logistic regression classifiers based on the two created data sets as well as the initial data set:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/anaconda3/lib/python3.6/site-packages/sklearn/utils/validation.py:752: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean squared error original data:  0.19180539207443528\n",
      "Mean squared error      dataset1:  0.2503222913513817\n",
      "Mean squared error      dataset2:  0.2978532593464492\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/anaconda3/lib/python3.6/site-packages/sklearn/utils/validation.py:752: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "/anaconda3/lib/python3.6/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/anaconda3/lib/python3.6/site-packages/sklearn/utils/validation.py:752: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'print(y_pred[0:10])\\nprint(y_pred[0:10])\\nprint(y_pred[0:10])'"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# your code\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "X = dataset.iloc[:,0:-1] #select all feature columns excluding label\n",
    "y_training_6c = y = pd.get_dummies(dataset['label'], drop_first=True) #drop_first to get one binary column representing of the two states in numbers\n",
    "\n",
    "#classifier for original dataset\n",
    "classifierLogistic = LogisticRegression()\n",
    "classifierLogistic.fit(X, y)\n",
    "y_predTraining_6c = y_pred = classifierLogistic.predict(X)\n",
    "\n",
    "#classifier for dataset1\n",
    "X = dataset1\n",
    "classifierLogistic1 = LogisticRegression()\n",
    "classifierLogistic1.fit(X, y)\n",
    "y_pred1 = classifierLogistic1.predict(X)\n",
    "\n",
    "#classifier for dataset2\n",
    "X = dataset2\n",
    "classifierLogistic2 = LogisticRegression()\n",
    "classifierLogistic2.fit(X, y)\n",
    "y_pred2 = classifierLogistic2.predict(X)\n",
    "\n",
    "#print the mean squared errors of each classifier\n",
    "print('Mean squared error original data: ', mean_squared_error(y, y_pred))\n",
    "print('Mean squared error      dataset1: ', mean_squared_error(y, y_pred1))\n",
    "print('Mean squared error      dataset2: ', mean_squared_error(y, y_pred2))\n",
    "\n",
    "#print the first 10 predicitons of each classifier\n",
    "\"\"\"print(y_pred[0:10])\n",
    "print(y_pred[0:10])\n",
    "print(y_pred[0:10])\"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "6. (d) Interpret the three resulting models and compare them. Which model do you recommend and why?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: \n",
    "\n",
    "Interpreting the mean squared error as loss function would mean that the classifier on the original dataset is the best, because it has the lowest error. Nevertheless, this evaluation is based on the training data and **not** on test data (simplified evaluation process as described in the lecture). In other words this \"good\" model could be overfitted and perform not good on the test data as on the training data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 7 (Neural Networks)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "7. (a) What are the inputs of your network?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: The inputs for the classification problem with a neural network are the 10 attributes (desriptive features) from the dataset (excluded *education* and the target *label*). It is not necessary to use all the 11 attributes (excluded target *label*), because *education* and *education-num* are redundant (compare question 1). They describe the same feature only in two syntactic ways. This means that in our case we can use only on of this attributes, in other words we use *education-num* for the Neural Network. Due to the fact that the input are numeric values, we have to encode the categorical attributes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "7. (b) What are the possible number of input pattern for your network (just including categorical attributes)?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: As input pattern all combinations of the 8 categorical attributes (*workclass, education-num, marital-status, occupation, relationship, race, sex, native-country*) are suitable.  \n",
    "This means that our Neural Network has got 10 inputs in the input layer in which each neuron represents one of 10 attributes (excluded *education* and the target *label*)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "7. (c.i) First, with default parameters and return the parameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "ename": "FileNotFoundError",
     "evalue": "[Errno 2] File b'adult.data-3.csv' does not exist: b'adult.data-3.csv'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mFileNotFoundError\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-2-5c7ec691fb14>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;31m#read data from csv\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 10\u001b[0;31m \u001b[0mdf_training\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mread_csv\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'adult.data-3.csv'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     11\u001b[0m \u001b[0mdf_test\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mread_csv\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'adult.test-3.csv'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/mnt/dev/anaconda3/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36mparser_f\u001b[0;34m(filepath_or_buffer, sep, delimiter, header, names, index_col, usecols, squeeze, prefix, mangle_dupe_cols, dtype, engine, converters, true_values, false_values, skipinitialspace, skiprows, skipfooter, nrows, na_values, keep_default_na, na_filter, verbose, skip_blank_lines, parse_dates, infer_datetime_format, keep_date_col, date_parser, dayfirst, iterator, chunksize, compression, thousands, decimal, lineterminator, quotechar, quoting, doublequote, escapechar, comment, encoding, dialect, tupleize_cols, error_bad_lines, warn_bad_lines, delim_whitespace, low_memory, memory_map, float_precision)\u001b[0m\n\u001b[1;32m    700\u001b[0m                     skip_blank_lines=skip_blank_lines)\n\u001b[1;32m    701\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 702\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0m_read\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfilepath_or_buffer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    703\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    704\u001b[0m     \u001b[0mparser_f\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__name__\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/mnt/dev/anaconda3/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36m_read\u001b[0;34m(filepath_or_buffer, kwds)\u001b[0m\n\u001b[1;32m    427\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    428\u001b[0m     \u001b[0;31m# Create the parser.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 429\u001b[0;31m     \u001b[0mparser\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTextFileReader\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfilepath_or_buffer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    430\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    431\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mchunksize\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0miterator\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/mnt/dev/anaconda3/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, f, engine, **kwds)\u001b[0m\n\u001b[1;32m    893\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moptions\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'has_index_names'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mkwds\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'has_index_names'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    894\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 895\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_engine\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mengine\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    896\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    897\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mclose\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/mnt/dev/anaconda3/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36m_make_engine\u001b[0;34m(self, engine)\u001b[0m\n\u001b[1;32m   1120\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_make_engine\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mengine\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'c'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1121\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mengine\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'c'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1122\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_engine\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mCParserWrapper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moptions\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1123\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1124\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mengine\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'python'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/mnt/dev/anaconda3/lib/python3.6/site-packages/pandas/io/parsers.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, src, **kwds)\u001b[0m\n\u001b[1;32m   1851\u001b[0m         \u001b[0mkwds\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'usecols'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0musecols\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1852\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1853\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_reader\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mparsers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTextReader\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1854\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munnamed_cols\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_reader\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munnamed_cols\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1855\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mpandas/_libs/parsers.pyx\u001b[0m in \u001b[0;36mpandas._libs.parsers.TextReader.__cinit__\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mpandas/_libs/parsers.pyx\u001b[0m in \u001b[0;36mpandas._libs.parsers.TextReader._setup_parser_source\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;31mFileNotFoundError\u001b[0m: [Errno 2] File b'adult.data-3.csv' does not exist: b'adult.data-3.csv'"
     ]
    }
   ],
   "source": [
    "# your code\n",
    "# no code from above for execution needed --- new section\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "import sklearn.metrics as metric\n",
    "\n",
    "#read data from csv\n",
    "df_training = pd.read_csv('adult.data-3.csv')\n",
    "df_test = pd.read_csv('adult.test-3.csv')\n",
    "\n",
    "X_training = df_training[['age','workclass','education-num','marital-status','occupation','relationship','race','sex','hours-per-week','native-country']]\n",
    "X_training = pd.get_dummies(X_training, columns = ['workclass','marital-status','occupation','relationship','race','sex','native-country'])\n",
    "y_training = df_training['label']\n",
    "\n",
    "X_test = df_test[['age','workclass','education-num','marital-status','occupation','relationship','race','sex','hours-per-week','native-country']]\n",
    "X_test = pd.get_dummies(X_test, columns = ['workclass','marital-status','occupation','relationship','race','sex','native-country'])\n",
    "y_test = df_test['label']\n",
    "\n",
    "#function for MLPClassifier with sklearn default values\n",
    "def mlp(layerTupel=(100), activationFunction='relu', params=False, save=False):\n",
    "    mlp = MLPClassifier(hidden_layer_sizes=layerTupel, activation=activationFunction)\n",
    "    mlp.fit(X_training, y_training)#training\n",
    "\n",
    "    y_predTraining = mlp.predict(X_training)  \n",
    "    y_predTest = mlp.predict(X_test)   \n",
    "\n",
    "    accuracyTraining = metric.accuracy_score(np.array(y_training).flatten(), np.array(y_predTraining).flatten(), normalize=True)\n",
    "    accuracyTest = metric.accuracy_score(np.array(y_test).flatten(), np.array(y_predTest).flatten(), normalize=True)\n",
    "    print('Shape: ', [coef.shape for coef in mlp.coefs_]) \n",
    "    print('Accuracy training =', accuracyTraining)                         \n",
    "    print('Accuracy test     =', accuracyTest,'\\n')\n",
    "   \n",
    "    if(params):\n",
    "        print('Parameters: ', mlp.get_params())\n",
    "    if(save):\n",
    "        global y_predTraining_7c, y_training_7c, y_predTest_7c, y_test_7c\n",
    "        y_predTraining_7c = y_predTraining\n",
    "        y_training_7c = y_training\n",
    "        y_predTest_7c = y_predTest\n",
    "        y_test_7c = y_test\n",
    "\n",
    "mlp(params=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "7. (c.ii) Second, try to find the optimized number of hidden layers and nodes. (Start with default number and then at least go with one number above and one number below the default) \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape:  [(88, 1)]\n",
      "Accuracy training = 0.8068493918502326\n",
      "Accuracy test     = 0.806707204717155 \n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:562: ConvergenceWarning: Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape:  [(88, 100), (100, 100), (100, 1)]\n",
      "Accuracy training = 0.8548848158735497\n",
      "Accuracy test     = 0.796265585651987 \n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:562: ConvergenceWarning: Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape:  [(88, 100), (100, 100), (100, 100), (100, 1)]\n",
      "Accuracy training = 0.8685051286362871\n",
      "Accuracy test     = 0.763589460106873 \n",
      "\n",
      "Shape:  [(88, 80), (80, 1)]\n",
      "Accuracy training = 0.8183958298301665\n",
      "Accuracy test     = 0.8212026288311529 \n",
      "\n",
      "Shape:  [(88, 88), (88, 1)]\n",
      "Accuracy training = 0.8234964407824673\n",
      "Accuracy test     = 0.8119280142497389 \n",
      "\n",
      "Shape:  [(88, 90), (90, 1)]\n",
      "Accuracy training = 0.8233843394428564\n",
      "Accuracy test     = 0.8018549229162828 \n",
      "\n",
      "Shape:  [(88, 99), (99, 1)]\n",
      "Accuracy training = 0.83156773723446\n",
      "Accuracy test     = 0.8004422332780542 \n",
      "\n",
      "Shape:  [(88, 110), (110, 1)]\n",
      "Accuracy training = 0.8230480354240233\n",
      "Accuracy test     = 0.8152447638351452 \n",
      "\n",
      "Shape:  [(88, 120), (120, 1)]\n",
      "Accuracy training = 0.825794518244493\n",
      "Accuracy test     = 0.7865610220502426 \n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:562: ConvergenceWarning: Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape:  [(88, 130), (130, 1)]\n",
      "Accuracy training = 0.832352446611737\n",
      "Accuracy test     = 0.8134635464652048 \n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:562: ConvergenceWarning: Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape:  [(88, 80), (80, 90), (90, 1)]\n",
      "Accuracy training = 0.852698839751135\n",
      "Accuracy test     = 0.7906148270990725 \n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:562: ConvergenceWarning: Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape:  [(88, 110), (110, 120), (120, 1)]\n",
      "Accuracy training = 0.8575191973544084\n",
      "Accuracy test     = 0.7706529082980161 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# your code for lower number of hidden layers\n",
    "mlp(())\n",
    "\n",
    "# your code for higher number of hidden layers\n",
    "mlp((100,100))\n",
    "mlp((100,100,100))  \n",
    "\n",
    "# your code for testing number of neurons in each layer\n",
    "mlp((80))\n",
    "mlp((88),save=True)\n",
    "mlp((90))\n",
    "mlp((99))\n",
    "mlp((110))\n",
    "mlp((120))\n",
    "mlp((130))\n",
    "mlp((80,90)) \n",
    "mlp((110,120)) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "7. (c.iii) Third, try to train your model with one linear activation function and one non linear activation function, name the functions and explain if there is any difference in your networks and why? (you can use evaluation metrics to show which activation function works better for this data set)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape:  [(88, 100), (100, 1)]\n",
      "Accuracy training = 0.806569138501205\n",
      "Accuracy test     = 0.8070757324488668 \n",
      "\n",
      "Shape:  [(88, 100), (100, 100), (100, 1)]\n",
      "Accuracy training = 0.7896418362199429\n",
      "Accuracy test     = 0.7430133284196302 \n",
      "\n",
      "Shape:  [(88, 88), (88, 1)]\n",
      "Accuracy training = 0.806569138501205\n",
      "Accuracy test     = 0.7943615257048093 \n",
      "\n",
      "Shape:  [(88, 88), (88, 88), (88, 88), (88, 1)]\n",
      "Accuracy training = 0.806569138501205\n",
      "Accuracy test     = 0.7938087341072416 \n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:562: ConvergenceWarning: Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape:  [(88, 100), (100, 1)]\n",
      "Accuracy training = 0.8263550249425481\n",
      "Accuracy test     = 0.8056630428106382 \n",
      "\n",
      "Shape:  [(88, 100), (100, 100), (100, 1)]\n",
      "Accuracy training = 0.8202455019337481\n",
      "Accuracy test     = 0.8073214176033413 \n",
      "\n",
      "Shape:  [(88, 88), (88, 1)]\n",
      "Accuracy training = 0.82545821422566\n",
      "Accuracy test     = 0.8056630428106382 \n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:562: ConvergenceWarning: Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    }
   ],
   "source": [
    "# your code for linear activation function\n",
    "mlp(activationFunction='identity')\n",
    "mlp((100,100), 'identity')\n",
    "mlp((88), 'identity')\n",
    "mlp((88,88,88), 'identity')\n",
    "\n",
    "# your code for non linear activation function\n",
    "mlp(activationFunction='logistic')\n",
    "mlp((100,100), 'logistic')\n",
    "mlp((88), 'logistic')\n",
    "\n",
    "# your code for comparing two networks \n",
    "#evaluation metrics are printed in the mlp-function for easier handling\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Parameters: *activationFunction*:\n",
    "- 'identity' - linear\n",
    "- 'logistic' - non-linear"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: In general, we can see that a non-linear activation function performs better in the metrics of accuracy. This is caused by the introduced non-linearity to the network, so that not only linear, moreover non-linear mappings from input to output can be generated. This means that we are not only limited to linearity."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "7. (c.iv) Which model do you recommend to be used as your classifier and why? (with respect to the number of hidden layer and activation function (linear or nonlinear))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: With respect to the accuray on the training data, I would recommend a neural network with a \"higher\" number of hidden layers and neurons in each layer. Additionally, a non-linear activation function to break down possible linearity in the network.  \n",
    "But, due to this recommendation, the probability of getting a overfitted model (difference of accuracy on training and test data) increases because the model is too precise - learns \"too much\". This leads to another approach of evaluation the models, e.g. with cross-validation, to counter to that overfitting behaviour."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Analysing the quality of the prediction models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 8 (Evaluation metrics)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "8. (a) Calculate the following metrics for the model 7. (c, i) and the logistic regression with all the initial attributes in question 6 (c):"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Confusion matrices on the training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix training data 7c: \n",
      "[[8329 1671]\n",
      " [1478 6363]]\n",
      "TN: 8329\n",
      "FP: 1671\n",
      "FN: 1478\n",
      "TP: 6363 \n",
      "\n",
      "Confusion matrix training data 6c: \n",
      "[[8134 1866]\n",
      " [1556 6285]]\n",
      "TN: 8134\n",
      "FP: 1866\n",
      "FN: 1556\n",
      "TP: 6285 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# your code for the confusion matrices on the training data\n",
    "# Necessary: Please run Q2 and Q6 before\n",
    "\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "cm_training_7c = confusion_matrix(np.array(y_training_7c), np.array(y_predTraining_7c))\n",
    "tn, fp, fn, tp = cm_training_7c.ravel()\n",
    "print('Confusion matrix training data 7c: ')\n",
    "print(cm_training_7c)\n",
    "print('TN:', tn)\n",
    "print('FP:', fp)\n",
    "print('FN:', fn)\n",
    "print('TP:', tp, '\\n')\n",
    "\n",
    "cm_training_6c = confusion_matrix(np.array(y_training_6c), np.array(y_predTraining_6c))\n",
    "tn, fp, fn, tp = cm_training_6c.ravel()\n",
    "print('Confusion matrix training data 6c: ')\n",
    "print(cm_training_6c)\n",
    "print('TN:', tn)\n",
    "print('FP:', fp)\n",
    "print('FN:', fn)\n",
    "print('TP:', tp, '\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Confusion matrices on the test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix test data 7c: \n",
      "[[10209  2226]\n",
      " [  836  3010]]\n",
      "TN: 10209\n",
      "FP: 2226\n",
      "FN: 836\n",
      "TP: 3010 \n",
      "\n",
      "Confusion matrix test data 6c: \n",
      "[[10064  2371]\n",
      " [  796  3050]]\n",
      "TN: 10064\n",
      "FP: 2371\n",
      "FN: 796\n",
      "TP: 3050 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# your code for the confusion matrices on the test data\n",
    "import pandas as pd\n",
    "\n",
    "cm_test_7c = confusion_matrix(np.array(y_test_7c), np.array(y_predTest_7c))\n",
    "tn, fp, fn, tp = cm_test_7c.ravel()\n",
    "print('Confusion matrix test data 7c: ')\n",
    "print(cm_test_7c)\n",
    "print('TN:', tn)\n",
    "print('FP:', fp)\n",
    "print('FN:', fn)\n",
    "print('TP:', tp, '\\n')\n",
    "\n",
    "y_predTest_6c = classifierLogistic.predict(X_test)\n",
    "y_test_6c = pd.get_dummies(y_test, drop_first=True)\n",
    "cm_test_6c = confusion_matrix(np.array(y_test_6c), np.array(y_predTest_6c))\n",
    "tn, fp, fn, tp = cm_test_6c.ravel()\n",
    "print('Confusion matrix test data 6c: ')\n",
    "print(cm_test_6c)\n",
    "print('TN:', tn)\n",
    "print('FP:', fp)\n",
    "print('FN:', fn)\n",
    "print('TP:', tp, '\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Precision, recall and F1-scores on the test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision test data 7c:  [0.92430964 0.57486631]\n",
      "Recall    test data 7c:  [0.82098914 0.78263131]\n",
      "F1-scores test data 7c:  [0.86959114 0.66284959] \n",
      "\n",
      "Precision test data 6c:  [0.9267035  0.56262682]\n",
      "Recall    test data 6c:  [0.80932851 0.79303172]\n",
      "F1-scores test data 6c:  [0.86404808 0.6582497 ]\n"
     ]
    }
   ],
   "source": [
    "# your code for precision, recall and F1-scores on the test data\n",
    "from sklearn.metrics import precision_score\n",
    "from sklearn.metrics import recall_score\n",
    "from sklearn.metrics import f1_score\n",
    "\n",
    "print('Precision test data 7c: ', precision_score(y_test_7c, y_predTest_7c, average=None))\n",
    "print('Recall    test data 7c: ', recall_score(y_test_7c, y_predTest_7c, average=None))\n",
    "print('F1-scores test data 7c: ', f1_score(y_test_7c, y_predTest_7c, average=None), '\\n')\n",
    "\n",
    "print('Precision test data 6c: ', precision_score(y_test_6c, y_predTest_6c, average=None))\n",
    "print('Recall    test data 6c: ', recall_score(y_test_6c, y_predTest_6c, average=None))\n",
    "print('F1-scores test data 6c: ', f1_score(y_test_6c, y_predTest_6c, average=None))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Accuracy scores on training and test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy training data 7c:  0.8234964407824673\n",
      "Accuracy test     data 7c:  0.8119280142497389 \n",
      "\n",
      "Accuracy training data 6c:  0.8081946079255647\n",
      "Accuracy test     data 6c:  0.8054787789447823\n"
     ]
    }
   ],
   "source": [
    "# your code for accuracy scores on training and test data\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "print('Accuracy training data 7c: ', accuracy_score(y_training_7c, y_predTraining_7c))\n",
    "print('Accuracy test     data 7c: ', accuracy_score(y_test_7c, y_predTest_7c), '\\n')\n",
    "\n",
    "print('Accuracy training data 6c: ', accuracy_score(y_training_6c, y_predTraining_6c))\n",
    "print('Accuracy test     data 6c: ', accuracy_score(y_test_6c, y_predTest_6c))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to turn numbers into insights, please comment on your findings. Motivate the answers to the following questions using the metrics and the findings in the questions 1 through 7."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "8. (b) What is, in your opinion, the best model? Motivate your answer with the correct performance metrics."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation:\n",
    "With help of the confusion matrix, we can detect an unequal distribution of the training data (imbalanced data) which indicates the usage of the F1-score, and not the precision score. In contrast, if False Negatives (FN) are associated with high costs, recall should be used. \n",
    "So, due to the fact that non of our labels \"*<50K*\" and \"*>50K*\" is rated to positive by default, the scores indicate both classes in a list. This means that there is no \"best\" model and it depends on the requirements which one is the most suitable. In our case the False Negatives (FN) should not be associated with high costs. Thus, we can compare the F1-score. Besides, we can say in general that the trained logistic regression model is always the same on the same training data. Moreover, in case of neural network it is not. This means that e.g. it is complicated to save the best model for comparing it in this question because we can not predict the \"best\" model and the values are changing all the time. Also the risk of overfitting (difference of accuracy on training and test data) is much higher at neural network compared to our regression model.\n",
    "Although all performance metrics distinguish in a very small amount, in my opinion I would use neural networks because in comparison the accuracy and the recall (if needed) is better. In addition, we know from the lecture that the quality of prediction of deep neural network is better than logistic regression (in case of a not-small data set, like in our case)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "8. (c) Does any model suffer from underfitting or overfitting? Motivate your answer with the correct performance metrics."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation:\n",
    "In our case the neural network model suffers more from overfitting (compare question 8b). But it seems, that in general both models does not suffer too much from overfitting, although there are higher outliers at neural network (compare resultion accuracies question 7).\n",
    "\n",
    "No model suffers from underfitting because non of them performs bad (compare accuracies) on training and test data (accuracy significant worse/ under the assumption that performing \"bad\" means less than 50% accuracy)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "8. (d) In the hypothesis that a separate validation set is not available, how would you proceed in evaluating a model? Refer to the evaluation techniques seen in the lectures."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Explanation: Since cross-validation is not possible due to the missing validation set, I would separate the data set to use 80% for training and 20% for testing, and perform the \"simple\" testing of the trained model."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}