Product Recommendation Dataset
Description
This dataset contains 1,000 market-basket transactions designed for association-rule mining and clustering experiments. It includes seven retail products: Bread, Cheese, Eggs, Milk, Yogurt, Chips, and Soda. The dataset is structured with transaction identifiers, product lists, and binary item-presence indicators.The dataset was generated using aggregate item-support statistics reported in a preliminary association-rule analysis. It is intended for experimental evaluation of the Apriori and FP-Growth algorithms, along with K-Means and K-Medoids clustering techniques. The dataset can be used to calculate and analyze association-rule metrics, including support, confidence, and lift, and to investigate product co-occurrence patterns.The reported aggregate statistics guided its generation; therefore, the original association rules are not guaranteed to be reproduced exactly. This dataset is intended for algorithm testing, methodological demonstrations, and reproducible experiments in market-basket analysis and data mining.The actual 1000 records were collected from the different shops
Files
Steps to reproduce
Steps to Reproduce Download the dataset in CSV or Excel format and load it into Python using Pandas. Inspect the dataset to identify transaction IDs, product names, and binary item-presence indicators. The seven products are Bread, Cheese, Eggs, Milk, Yogurt, Chips, and Soda. Preprocess the transaction data by checking missing values, duplicate records, and transaction consistency. Use the binary item-presence columns or convert the product lists into a transaction matrix. Apply the Apriori algorithm to identify frequent itemsets using a selected minimum support threshold. Generate association rules and calculate support, confidence, and lift. Apply the FP-Growth algorithm to identify frequent itemsets and generate association rules using the same support threshold for comparison with Apriori. Apply K-Means and K-Medoids clustering to the binary transaction features. Select the number of clusters and evaluate the resulting cluster assignments. Perform association-rule mining within the appropriate clusters using Apriori and FP-Growth. Calculate support, confidence, and lift for the resulting rules. Compare the algorithms and clustering combinations based on the number of frequent itemsets, association rules, computational time, and rule metrics. Record the software versions, parameter settings, random seeds, and evaluation procedures to support reproducibility.Recommended software: Python 3, Pandas, NumPy, mlxtend for Apriori and FP-Growth, and scikit-learn-extra or another documented implementation for K-Medoids.