Market Basket Analysis: Support, Confidence, Lift & Cross-Selling
Item and pair support from 0/1 basket rows, a co-occurrence matrix with SUMPRODUCT and INDEX/MATCH, confidence, lift, leverage and conviction, filtering and ranking rules, the Apriori pruning idea on 3-item sets and a cross-selling plan with expected margin uplift: 8 graded labs on one fictional grocery store.
The first lesson is free — no account needed.
About this course
A market basket analysis course that mines the shopping baskets of one fictional neighbourhood grocery store for association rules in a spreadsheet, for retail and e-commerce analysts, category managers, data science students and anyone who wants to know what a "frequently bought together" box is built on. You turn 24 baskets into 0/1 rows and calculate item support and the average basket size, count pair support with SUMPRODUCT and fill a whole co-occurrence matrix with one formula, look up counts with INDEX and MATCH to calculate the confidence of a rule in both directions, label rules by lift as complements, independent or substitutes, add leverage and conviction, filter 14 candidate rules by minimum support and confidence and rank them by lift, prune 3-item sets the Apriori way and turn the best rules into a cross-selling plan with expected margin uplift. The first lab is free.
Lessons
- Baskets as 0/1 Rows: Item Frequency and SupportFree
- Pair Support with SUMPRODUCT: the Co-Occurrence Matrix Pro
- Confidence A → B: How Often B Follows A Pro
- Lift: Is the Rule Better than Chance? Pro
- Leverage and Conviction: Size and Direction of a Rule Pro
- Filtering Rules by Minimum Support and Confidence, Ranking by Lift Pro
- The Apriori Idea: Pruning 3-Item Sets Before You Count Them Pro
- From Rules to Cross-Selling: Expected Uplift per Recommendation Pro