Mumbai House Price Prediction

My first Data Science and Machine Learning project! In this project, I've developed a model to predict house prices in Mumbai using various features. The dataset used for this project was obtained from Kaggle. You can find the dataset here.

Learning and Growth

I want to emphasize that I am still learning and growing as a data scientist. This project reflects my current understanding, and I'm excited to see how my skills will evolve with each new project I undertake. Constructive feedback and suggestions are always welcome as they contribute to my learning process.

Project Overview

This project is about predicting house prices in Mumbai, a city with a dynamic real estate market. The goal was to create a machine learning model that could predict house prices based on different attributes like square footage, number of bedrooms, location, etc.

Data Exploration and Cleaning

• I started by exploring the dataset obtained from Kaggle.
• The dataset had a mix of good and not-so-good quality data.
• I performed data cleaning, which involved handling missing values and converting the price column into proper numerical values.
• I used data visualization techniques such as charts and graphs to gain insights into the data distribution and relationships.

Feature Engineering and Preprocessing

• I dropped unnecessary columns that wouldn't contribute to the prediction.
• Outliers were detected and handled to improve the model's robustness.
• I used one-hot encoding to convert categorical variables, especially the 'location' feature, into numerical values for model training.

Model Selection and Training

• I evaluated three different regression models: Linear Regression, Random Forest Regressor, and Decision Tree Regressor.
• To find the best model, I utilized GridSearchCV for hyperparameter tuning.
• Based on the results, the Random Forest Regressor showed the best performance, so I selected it as the final model.

Model Deployment

• After selecting the Random Forest Regressor as the best model, I trained it on the cleaned dataset.
• I saved the trained model as a pickle file, which can be used for making predictions without retraining the model every time.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Happy coding!