ADS-overview

1. Visualise POIs for different locations & tags (e.g., tourism POIs around Cambridge train station).
2. Use a heatmap for visualising the OSM feature correlation.
3. Apply PCA on the correlation matrix, and visualise relative distances between features with an interactive 3D plot (users can change elevation & azimuth).

1. Select an optimal bounding box and date range automatically.
2. Transform prices: a constrained domain to an unconstrained one using log.
3. Preprocess data: handle missing and invalid data, merge with OSM features, apply one-hot encoding, and preserve only quantitative data.
4. Apply cross-validation with RMSE for evaluating model performance.
5. Linear models include: RidgeCV, BayesianRidge, and TweedieRegressor.
6. Predict prices and report model performance (warn if the model is poor).

Address

Assess

1. Automatically download data and upload files to the database (use a progress bar whenever possible).
2. APIs for joining tables, converting queried results into DataFrames, and getting POI data from OSM.

Access

Address

Assess

Fynesse

House Price Prediction

Access