In this project, we'll predict future house prices. We'll use data from the Federal Reserve, along with house price data from Zillow. We'll merge and combine this data, then use it to train a random forest model. The model will predict if house prices will increase or decrease in the future. We'll measure error using backtesting, then improve our model with new predictors.
This project can be customized to predict house prices in your metro area if you live in the US.
Project Steps
- Load in data
- Clean and merge data
- Create an initial machine learning model and estimate accuracy
- Improve the accuracy of the model
- Run diagnostics to figure out how we can improve
You can find the code for this project here
File overview:
prices.ipynb- a Jupyter notebook that contains all of the code.
To follow this project, please install the following locally:
- JupyerLab
- Python 3.8+
- Python packages
- pandas
- yfinance
- scikit-learn
You'll need to download a few csv files to run this project. These files are included in this repo, but you may want newer versions:
- Federal reserve data
- CPI dataset - CPIAUCSL.csv
- Rental vacancy rate - RRVRUSQ156N.csv
- Mortgage interest rates - MORTGAGE30US.csv
- Zillow data
- ZHVI (raw, weekly) - Metro_zhvi_uc_sfrcondo_tier_0.33_0.67_month.csv
- Median sale price (raw, all homes, weekly) - Metro_median_sale_price_uc_sfrcondo_week.csv