rbnuria / Purchase100-dataset

A repo to download and preprocess the Purchase100 dataset extracted from Kaggle: Acquire Valued Shoppers Challenge

Purchase100 dataset

A repo to preprocess the Purchase100 dataset extracted from Kaggle: Acquire Valued Shoppers Challenge

Download preprocessed dataset

• purchase100.npz

The authenticity of the downloaded files can be checked with the following md5 hashes

• purchase100.npz : 0d7538b9806e7ee622e1a252585e7768

Steps to preprocess the dataset

1. Download the transactions.csv.gz file from https://www.kaggle.com/c/acquire-valued-shoppers-challenge/data
2. Run preprocess_dataset.py <path_to_transactions.csv.gz> to generate purchase100.npz

How to use it

• You can import them in your code using numpy.load function

   data = np.load('./purchase100.npz')
   features = data['features']
   labels = data['labels']

Requirements

This work is tested with Python 3.8.5.

The requirements.txt file is automatically generated with pipreqs.

References

The code in this repo is based on the preprocessing scripts given in https://github.com/bargavj/EvaluatingDPML