Instacart Customer Reorder Prediction
- This is my attempt at Instacart Market Basket Analysis competition hosted by Kaggle.
- Detailed description of the competition and its goal can be found in the link above.
01. Data Integrity Check
- Check to see if there are any possible missing values and mismatch between the datasets as the dataset is spread accross multiple files in different formats.
02. Exploratory Data Analysis
- Gain some insight into the given dataset.
03. Prediction
- A binary decision tree model (BDT) is used.
- The goal here was not to get the highest possible score, but to use a small number of features with a fast computing model and get a score that is reasonably close to the score attained by a more robust gradient boosting method (in particular, XGBoost).
- Compared to the base XGBoost model used by most participants, which gives the score of 0.38 after several hours of computation, it was possible to attain a score of 0.37 under a minute with BDT.