This project aims to predict house prices based on various features, using machine learning techniques. It involves data cleaning and preprocessing, feature engineering, model selection and training and evaluation to achieve the best predictive performance.

The dataset used is the California Housing Prices dataset from Kaggle. It contains information about various houses in California, including features like:

longitude
latitude
housing_median_age
total_rooms
total_bedrooms
population
households
median_income
median_house_value
ocean_proximity

Handling Missing Values: Removed rows with missing values to ensure data integrity.
Encoding Categorical Variables: Used one-hot encoding for the ocean_proximity feature.
Feature Engineering: Created new features eg bedroom_ratio and household_room.
Data Scaling: Scaled the features for better model performance.

Linear Regression: Trained a simple Linear Regression model, achieving a score of 0.66.
Random Forest Regressor: Trained a more complex Random Forest model, achieving a score of 0.76.
Hyperparameter Tuning: Used GridSearchCV to tune the Random Forest model, confirming the best parameters and maintaining the score at 0.77.

Evaluated the models using the test set to compare their performance. The evaluation metric used was the model score.

Linear Regression: Score of 0.66.
Random Forest Regressor: Score of 0.76.
Optimized Random Forest: Best parameters (n_estimators=3, max_features=4) with a score of 0.77.

Correlation Matrix: Visualized the correlation between different features.
Scatter Plot: Plotted latitude vs longitude with house prices.

This project provided valuable insights into the importance of data preprocessing, feature engineering, and model optimization in machine learning. The Random Forest model showed significant improvement over Linear Regression, and hyperparameter tuning further enhanced the model’s performance.