Data for this challenge has been collected over 11 months during Alvin’s Beta release. If a user is the first user to purchase at a merchant, the app asks the user to manually classify the merchant. The next user to purchase at that merchant has the opportunity to confirm the suggestion or enter a new categorisation.
Alvin currently registers user transaction data via MPESA SMS receipts, but some users classify their purchases manually.
There are ~400 purchases in train and ~600 in test. These ~1000 transactions have been verified by Alvin as the correctly classified. There are ~10 000 unverified transactions in the unvervified file, these are purchases users have classified themselves and Alvin has not checked.
Note there are 13 classes. If your model does not predict all classes you will need to manually add the missing columns filled with 0.
The objective of this challenge is to create a machine learning algorithm that classifies each purchase into one of 13 different categories.
This shows the submission format for this competition. The order of the rows does not matter, but the names of the Transactiopn_ID must be correct.
Contains the target. This is the dataset that you will use to train your model.
Resembles Train.csv but without the target-related columns. This is the dataset on which you will apply your model to.
These are purchases users have classified themselves and Alvin has not checked