add numpy exercises

Author: bob7783

numpy_class/exercises/ex1.py

@@ -0,0 +1,29 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import matplotlib.pyplot as plt
+
+A = np.array([
+  [0.3, 0.6, 0.1],
+  [0.5, 0.2, 0.3],
+  [0.4, 0.1, 0.5]])
+
+v = np.ones(3) / 3
+
+num_iters = 25
+distances = np.zeros(num_iters)
+for i in range(num_iters):
+  v2 = v.dot(A)
+  d = np.linalg.norm(v2 - v)
+  distances[i] = d
+  v = v2
+
+plt.plot(distances)
+plt.show()

numpy_class/exercises/ex2.py

@@ -0,0 +1,29 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import matplotlib.pyplot as plt
+
+def sampleY(n=1000):
+  # draw n samples from uniform dist.
+  X = np.random.random(n)
+  Y = X.sum()
+  return Y
+
+
+# now draw N Y's
+N = 1000
+Y_samples = np.zeros(N)
+for i in range(N):
+  Y_samples[i] = sampleY()
+
+
+# now plot the Y_samples
+plt.hist(Y_samples, bins=20)
+plt.show()

numpy_class/exercises/ex3.py

@@ -0,0 +1,34 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+
+
+# load in the data
+df = pd.read_csv('../../large_files/train.csv')
+data = df.values
+X = data[:, 1:] # images
+Y = data[:, 0] # labels
+
+# loop through each label
+for k in range(10):
+  Xk = X[Y == k]
+
+  # mean image
+  Mk = Xk.mean(axis=0)
+
+  # reshape into an image
+  im = Mk.reshape(28, 28)
+
+  # plot the image
+  plt.imshow(im, cmap='gray')
+  plt.title("Label: %s" % k)
+  plt.show()

numpy_class/exercises/ex4.py

@@ -0,0 +1,54 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+
+
+# load in the data
+df = pd.read_csv('../../large_files/train.csv')
+data = df.values
+
+# shuffle the images
+np.random.shuffle(data)
+
+X = data[:, 1:] # images
+Y = data[:, 0] # labels
+
+
+# define rotate functions
+def rotate1(im):
+  return np.rot90(im, 3)
+
+def rotate2(im):
+  H, W = im.shape
+  im2 = np.zeros((W, H))
+  for i in range(H):
+    for j in range(W):
+      im2[j,H - i - 1] = im[i,j]
+  return im2
+
+
+for i in range(X.shape[0]):
+  # get the image
+  im = X[i].reshape(28, 28)
+
+  # flip the image
+  # im = rotate1(im)
+  im = rotate2(im)
+
+  # plot the image
+  plt.imshow(im, cmap='gray')
+  plt.title("Label: %s" % Y[i])
+  plt.show()
+
+  ans = input("Continue? [Y/n]: ")
+  if ans and ans[0].lower() == 'n':
+    break

numpy_class/exercises/ex5.py

@@ -0,0 +1,60 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import matplotlib.pyplot as plt
+
+
+
+def is_symmetric1(A):
+  return np.all(A == A.T)
+
+
+def is_symmetric2(A):
+  rows, cols = A.shape
+  if rows != cols:
+    return False
+
+  for i in range(rows):
+    for j in range(cols):
+      if A[i,j] != A[j,i]:
+        return False
+
+  return True
+
+
+def check(A, b):
+  print("Testing:", A)
+  assert(is_symmetric1(A) == b)
+  assert(is_symmetric2(A) == b)
+
+
+# test the functions
+A = np.zeros((3, 3))
+check(A, True)
+
+A = np.eye(3)
+check(A, True)
+
+A = np.random.randn(3, 2)
+A = A.dot(A.T)
+check(A, True)
+
+A = np.array([[1, 2, 3], [2, 4, 5], [3, 5, 6]])
+check(A, True)
+
+A = np.random.randn(3, 2)
+check(A, False)
+
+A = np.random.randn(3, 3)
+check(A, False)
+
+A = np.arange(9).reshape(3, 3)
+check(A, False)
+

numpy_class/exercises/ex6.py

@@ -0,0 +1,25 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import matplotlib.pyplot as plt
+
+
+# generate unlabeled data
+N = 2000
+X = np.random.random((N, 2))*2 - 1
+
+# generate labels
+Y = np.zeros(N)
+Y[(X[:,0] < 0) & (X[:,1] > 0)] = 1
+Y[(X[:,0] > 0) & (X[:,1] < 0)] = 1
+
+# plot it
+plt.scatter(X[:,0], X[:,1], c=Y)
+plt.show()

numpy_class/exercises/ex7.py

@@ -0,0 +1,34 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import matplotlib.pyplot as plt
+
+def get_donut():
+  N = 2000
+  R_inner = 5
+  R_outer = 10
+
+  # distance from origin is radius + random normal
+  # angle theta is uniformly distributed between (0, 2pi)
+  R1 = np.random.randn(N//2) + R_inner
+  theta = 2*np.pi*np.random.random(N//2)
+  X_inner = np.concatenate([[R1 * np.cos(theta)], [R1 * np.sin(theta)]]).T
+
+  R2 = np.random.randn(N//2) + R_outer
+  theta = 2*np.pi*np.random.random(N//2)
+  X_outer = np.concatenate([[R2 * np.cos(theta)], [R2 * np.sin(theta)]]).T
+
+  X = np.concatenate([ X_inner, X_outer ])
+  Y = np.array([0]*(N//2) + [1]*(N//2))
+  return X, Y
+
+X, Y = get_donut()
+plt.scatter(X[:,0], X[:,1], c=Y)
+plt.show()

numpy_class/exercises/ex8.py

@@ -0,0 +1,50 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import matplotlib.pyplot as plt
+
+def get_spiral():
+  # Idea: radius -> low...high
+  #           (don't start at 0, otherwise points will be "mushed" at origin)
+  #       angle = low...high proportional to radius
+  #               [0, 2pi/6, 4pi/6, ..., 10pi/6] --> [pi/2, pi/3 + pi/2, ..., ]
+  # x = rcos(theta), y = rsin(theta) as usual
+
+  radius = np.linspace(1, 10, 100)
+  thetas = np.empty((6, 100))
+  for i in range(6):
+      start_angle = np.pi*i / 3.0
+      end_angle = start_angle + np.pi / 2
+      points = np.linspace(start_angle, end_angle, 100)
+      thetas[i] = points
+
+  # convert into cartesian coordinates
+  x1 = np.empty((6, 100))
+  x2 = np.empty((6, 100))
+  for i in range(6):
+      x1[i] = radius * np.cos(thetas[i])
+      x2[i] = radius * np.sin(thetas[i])
+
+  # inputs
+  X = np.empty((600, 2))
+  X[:,0] = x1.flatten()
+  X[:,1] = x2.flatten()
+
+  # add noise
+  X += np.random.randn(600, 2)*0.5
+
+  # targets
+  Y = np.array([0]*100 + [1]*100 + [0]*100 + [1]*100 + [0]*100 + [1]*100)
+  return X, Y
+
+
+X, Y = get_spiral()
+plt.scatter(X[:,0], X[:,1], c=Y)
+plt.show()

numpy_class/exercises/ex9.py

@@ -0,0 +1,28 @@
+# https://deeplearningcourses.com/c/deep-learning-prerequisites-the-numpy-stack-in-python
+# https://www.udemy.com/deep-learning-prerequisites-the-numpy-stack-in-python
+
+from __future__ import print_function, division
+from future.utils import iteritems
+from builtins import range, input
+# Note: you may need to update your version of future
+# sudo pip install -U future
+
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+
+
+from ex8 import get_spiral
+
+# get the data
+X, Y = get_spiral()
+
+# combine the data into one array
+# data to be concatenated must have same # of dimensions
+# e.g. N x D and N x 1
+# not  N x D and N
+data = np.concatenate((X, np.expand_dims(Y, 1)), axis=1)
+
+df = pd.DataFrame(data)
+df.columns = ['x1', 'x2', 'y']
+df.to_csv('mydata.csv', index=False)