/recom-day-2020-challenge

Primary LanguageJupyter Notebook

Retail recommendation challenge

The Data

The data for this challenge is taken from real orders, real people made on the jul.co.il website.

In efforts to protect customer privacy, personally identifiable information had been removed, and the data had been resampled.

Software requirements

  1. Git
  2. Python 3.6+, with pandas, and jupyter installed - we recommend anaconda
  3. A browser with internent connectivity

The goal

jul_train.csv contains a sample of user orders up to a certain time.

All of these users made at least one more order, your goal is predict at least one of the products in their next order.

These predictions would be regarded as a product recommendation in this context.

Submitting your product recommendation

  1. Clone this repo
  2. Launch jupyter notebook and browse
  3. Run submission.ipynb and follow the instructions
  4. Once submitted, the score should be listed on the leaderboard

The baseline

Top achievers on the leaderboard can score 40% and above. In order to pass the test, you are required to score 20% or above.