Automate product recommendation systems based on sales trends
Better meet demand for products based on products purchased
Improve buying experience for customers by showing products they’re likely to be interested in
Market basket analysis reveals which items buyers purchase together. Retailers use market basket analysis to understand the best way to co-locate products in both physical and digital stores. It shows them how to cross-sell and up-sell items that customers often put into their shopping carts at the same time.
It’s not difficult to infer relationships between a small number of products (itemset) and a small number of transactions. But as the quantity of products increases, it becomes necessary to study a quickly growing number of transactions to arrive at meaningful, statistically supported conclusions. And as promotions and recommendations come into play, data points from other sources become important. Spreadsheet models don’t suffice for high-volume retailers trying to build associations among hundreds or thousands of products.
Market basket analysis is an analytic approach that includes variants like cross-selling and next-product analysis. It aims to describe the relationship between an “if” (antecedent) item and a “then” (consequent) item.
This analysis measures the frequency of the itemsets identified among all transactions, as well as the strength of the associations among those items. Analysts can apply metrics like support (proportion of transactions showing association), confidence (probability of purchasing the consequent, given the antecedent), and lift (strength of association between antecedent and consequent). With the Apriori and Eclat algorithms, they can apply association rule mining and pruning to use computational power efficiently and calculate metrics for many potential itemsets.
Alteryx Designer includes tools purpose-built for market basket analysis, including options for calculating support, confidence, and lift. All that’s needed is a dataset of transactions with the items included in each. With just two tools, analysts can generate association rules and view visualizations of those associations, including a network diagram displaying their relationships. Once the market basket analysis is complete, the results can be viewed in dash-boarding tools such as Tableau.
1 – Data Access
Connect data sources containing sales records
2 – Data Prep
Apply rules algorithm to create product associations
3 – Automated Results
View results and export into Tableau