Linear regression for house prices

What this project does

This beginner project teaches a simple linear regression pipeline for predicting numeric house sale prices from a small set of property features. You will build a synthetic but realistic toy dataset (floor area, location, age, property type), fit a scikit-learn LinearRegression baseline, and evaluate it honestly on a held-out test set using MAE and RMSE. The aim is to illustrate the difference between a fitted line and a reliable out-of-sample price estimate, and to show how a median baseline compares to the trained model.

Data

The dataset is synthetic and generated inside the code using numpy. It is a modest-sized simulated housing dataset where price is a linear function of floor area with added effects for location, age and property type plus noise. This represents a classroom stand-in for real market data; it is not real transaction data.

How to build it

1. Install the required packages (see requirements). Create a virtual environment if you prefer. Example: python -m venv venv && source venv/bin/activate (or venv\Scripts\activate on Windows).
2. Install dependencies: pip install -r requirements.txt
3. Run the project module: python project.py to print evaluation metrics for the held-out test set.
4. Run tests: pytest -q to run the automated checks and see the pipeline validated.

How to run

Run the main script directly: python project.py. It prints a JSON object with the main evaluation metrics and simple diagnostics.

How to test

Run: pytest -q. The test suite checks that the run function returns the expected keys, that metrics lie in reasonable ranges, that a conservative performance floor is met, and that results are deterministic for the same seed.

Results

Measured by running python project.py (seed 42) on 2026-10-23; the tests in test_project.py assert the floor.

Metric	Value
test_mae	7470.5067
test_rmse	9425.9956
test_r2	0.9991
baseline_mae	253735.2591
n_train	900
n_test	300

Read the test_r2 value with care: the data here is small and synthetic with a planted, learnable structure, so a near-perfect score shows the pipeline works, not that the method would reach this on real data. See Limitations.

Limitations

This is a small synthetic dataset intended for teaching. Real housing markets have many more relevant predictors (detailed location, condition, supply/demand, transaction timing, policy effects) and more complex, non-linear relationships. A simple linear model on synthetic data is useful for pedagogy but not for production pricing decisions.

Next steps

Possible extensions include: adding more realistic features (number of rooms, bathrooms, proximity to transport), using regularisation (Ridge/Lasso), exploring non-linear models (tree-based methods), and validating on real transaction data with proper geographic splits and temporal holdouts.

---

Full case study, build steps and evaluation checks: Linear regression model for predicting house prices on networkershome.com

Built by Networkers Home · Learn the skills in the AI Engineer Course · Part of 100 AI projects with code