This script identifies association rules based on existing records of buyer’s transactions at a supermarket. The input is a whitespace delimited data file, where each row records the SKU numbers of items in a single transaction. The script finds the most common items sold together. The set size for the common items is defined in the variable item_set_size, and a parameter sigma defines the minimum number of occurances of each set to return. Default values are 3 for item_set_size and 4 for sigma.
From the command line, run the script on the default dataset (retail_25k.dat) and using the default values for item_set_size = 3 and sigma = 4 using the command
python supermarket_optimization.py
An alternate dataset and different values for the model parameters can be specified by adding arguments (the order must be filename, item_set_size, sigma), such as
python supermarket_optimization.py retail_25k.dat 3 4
See the Python script for commented code and the Jupyter notebook for expected output.
Running the script on the retail_25k.dat dataset with item_set_size = 3 and sigma = 4 results in the file output.csv.