🏡 Multiple Linear Regression – California Housing Dataset

This project demonstrates Multiple Linear Regression using the popular California Housing dataset from sklearn.datasets. It explores feature relationships, evaluates model performance using multiple metrics, and finally prepares the model for deployment using Pickling.

---

📌 Project Overview

Multiple Linear Regression helps us model the relationship between one dependent variable (target) and multiple independent variables (features). In this project, we aim to predict housing prices based on various features from the California Housing dataset.

---

🗃️ Dataset Details

• 📦 Source: sklearn.datasets.fetch_california_housing
• 🧮 Samples: 20,000+
• 🔢 Features: 8 numerical features
• 🎯 Target: Price (Median House Value)

---

🧠 Workflow

1. Load Dataset & Create DataFrame

   • Loaded using fetch_california_housing()
   • Converted to a pandas DataFrame

2. Exploratory Data Analysis

   • Used seaborn.pairplot() to visualize relationships
   • Created a heatmap to observe feature correlations

3. Data Preparation

   • Split into train/test sets using train_test_split
   • Standardized features using StandardScaler

4. Model Building

   • Trained a Multiple Linear Regression model using LinearRegression from scikit-learn

5. Model Evaluation

   • Evaluated with:
     • Mean Squared Error (MSE)
     • Mean Absolute Error (MAE)
     • Root Mean Squared Error (RMSE)
     • R² Score
     • Adjusted R² Score

6. Assumptions & Residual Analysis

   • Plotted residuals using:
     • seaborn.distplot() to check normality
     • Scatter plot of residuals vs predictions to check homoscedasticity
   • Found that model accuracy could be improved; performance wasn't optimal

7. Model Deployment Prep

   • Exported the trained model using Pickling (pickle.dump)
   • Discussed its usage in cloud-based inference pipelines

---

📊 Libraries Used

Library	Purpose
pandas	Data handling
numpy	Numerical computation
seaborn	Visualization (pairplot, heatmap)
matplotlib	Plotting
sklearn	Dataset loading, ML models, metrics
pickle	Model serialization

---

📈 Metrics Used

• 📉 MSE – Mean Squared Error
• 📉 MAE – Mean Absolute Error
• 📉 RMSE – Root Mean Squared Error
• 📈 R² Score – Goodness of fit
• 📈 Adjusted R² – R² adjusted for number of features

---

🗃️ Project Structure

File Name	Description
Multiple_Linear_Regression.ipynb	Full model implementation and evaluation
README.md	Project documentation (this file)
model.pkl	Serialized (pickled) trained model

---

🚀 How to Run the Project

1. Clone the Repository

   git clone https://github.com/YourUsername/Multiple-Linear-Regression-California.git
   cd Multiple-Linear-Regression-California
   

2. Install required libraries

   pip install pandas numpy matplotlib seaborn scikit-learn
   

3. Launch Jupyter Notebook

   jupyter notebook
   

4. Open ipynb files and run through the cells.

☁️ Model Deployment Tip

To deploy this model on the cloud:

• Load the model.pkl file in your API/backend

• Use libraries like Flask, FastAPI, or cloud services like AWS Lambda / Azure Functions

• Standardize incoming input data exactly as done before training

• Perform prediction using:

  import pickle
  model = pickle.load(open("model.pkl", "rb"))
  prediction = model.predict(new_scaled_data)
  

👩‍💻 Author

Maitri Prabhu

GitHub: Mai3Prabhu