Multi-class-Classification-Logistic-Regression

The notebook is a complete hands-on lab for performing multi-class classification using a real-world dataset related to obesity levels. Here's a structured breakdown of its content:

---

🔍 Main Purpose

To demonstrate how to implement multi-class classification strategies in Python using scikit-learn on a labeled dataset about obesity.

---

📂 Dataset Used

• File: https://www.kaggle.com/datasets/ezzaldeenesmail/obesitydataset-raw-and-data-sinthetic
• Loaded with: pandas.read_csv()
• Target column: NObeyesdad (which represents obesity categories)

---

📌 Notebook Structure

1. Setup and Imports

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pylab as plt
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.multiclass import OneVsOneClassifier
from sklearn.metrics import accuracy_score

These are the main tools used for:

• Data processing
• Visualization
• Training and evaluating classification models

---

2. Data Loading & Exploration

• Load the dataset into a DataFrame
• Display first few records (data.head())
• Visualize target variable distribution:

sns.countplot(y='NObeyesdad', data=data)

---

3. Preprocessing

• Apply One-Hot Encoding for categorical variables
• Use StandardScaler to normalize numerical features
• Split data into training and testing sets

---

4. Modeling

• Implements a Logistic Regression model using:
  • One-vs-Rest (OvR) strategy
  • One-vs-One (OvO) strategy

# Example:
model = LogisticRegression()
ovo = OneVsOneClassifier(model)
ovo.fit(X_train, y_train)

---

5. Evaluation

• Evaluate model performance using accuracy_score
• Compare results from OvO and OvR classifiers

---

6. Conclusion & Summary

• Analyzes which strategy works better for the dataset
• Shows final accuracy and insights

---

Would you like a copy of the dataset used, a visual chart of the workflow, or a restructured version of the notebook for learning purposes?