ML-Course-Project

Real-estate price prediction comparison

House Price Prediction

This repository contains a project for predicting house prices using various regression models, including Linear Regression, Ridge Regression, Lasso Regression, and a Neural Network model. Additionally, a meta-model is used to combine the predictions from the individual regression models to improve overall performance.

Dataset

The dataset used for this project is train.csv. This dataset includes various features related to houses, and the target variable is SalePrice.

Project Overview

1. Data Cleaning and Preprocessing

   • Missing values are handled by dropping columns with a high percentage of missing values and filling remaining missing values with column means.
   • Categorical variables are transformed using one-hot encoding.
   • The dataset is split into training and testing sets.

2. Modeling

   • Linear Regression: A basic linear regression model is trained and evaluated.
   • Ridge Regression: A ridge regression model is trained, and hyperparameters are tuned using GridSearchCV.
   • Lasso Regression: A lasso regression model is trained, and hyperparameters are tuned using GridSearchCV.
   • Neural Network: A deep neural network model with multiple hidden layers is trained.
   • Meta-Model: Predictions from the three regression models are stacked and used as input features for a meta-model, which is a simple neural network.

3. Evaluation

   • Models are evaluated using Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE).
   • Results are visualized using scatter plots and loss curves.

Dependencies

• pandas
• numpy
• matplotlib
• seaborn
• scikit-learn
• tensorflow