House Price Forecasting

This project aims to forecast house prices using a time series analysis approach. It includes data preprocessing and forecasting steps using the SARIMAX model.

Project Structure

1. Data Preprocessing Pipeline

   • Cleans and prepares the data for analysis.

2. Forecasting

   • Uses the SARIMAX model to predict future house prices for different regions.

Preprocessing Steps

1. Missing Data Handling

   • Removes rows with more than 50% missing values.
   • Uses forward-fill, backward-fill and linear interpolation for numeric columns.

2. Duplicate Handling

   • Removes duplicate rows from the dataset.

3. Output

   • Saves the processed data to processed_house_prices.csv.

Forecasting Steps

1. Data Loading

   • Loads processed data from processed_house_prices.csv.

2. Data Transformation

   • Converts data to a long format suitable for time series analysis.

3. Modeling

   • Fits a SARIMAX model for each region to forecast house prices.

4. Output

   • Saves the forecasted house prices to house_price_forecast.csv with values rounded to 4 decimal places.

Combine Steps

1. Data Loading

   • Loads processed data from processed_house_prices.csv and forecasted data from house_price_forecast.csv

2. Data Transformation

   • Transpose the forecasted data.

3. Output

   • Saves the combined data to combined_data.csv.

Usage

1. Activate the virtual environment:

   # Windows
   myenv\Scripts\activate
   
   # Unix/MacOS
   source myenv/bin/activate   

2. Install dependencies:

   pip install -r requirements.txt   

3. Run the preprocessing script:

   python data_preprocessing.py   

4. Run the forecasting script:

   python forecast_house_price.py   

5. Run the combine script:

   python combine_data.py   

Input/Output

• Input: data.csv
• Output: processed_house_prices.csv
• Output: house_price_forecast.csv
• Output: combined_data.csv

Requirements

See requirements.txt for package dependencies.

Notes

• Ensure that the input data is properly formatted and preprocessed before running the forecasting script.
• The forecasting model is set to predict the next 5 months of house prices for each region.