Very quick sketch of the code used for Institut Louis Bachelier Data Chalenge 2022.
This repo contains the following notebooks:
The notebook I used for EDA, visualizing the target variable, missing values, showing a few plots.
Main notebook, used to generate a prediction. It gathers feature engineering part and modeling part. It also loads external data and images related features.
Generates images embedding using three backbones: ResNet, EfficientNet and Inception.
Uses images embeddings to make a prediction on the houses price. This prediction is then used in the model to increase overall performance.