THE RESULT
What the data shows
The largest computed metric is 7.994 for 20728 + 22384; the smallest is 0.7298 for 85099B + POST. Metric: lift (Lift).
Product search is restricted to 30 popular codes; require at least 50 co-purchase invoices. The denominator includes all positive-sales invoices.
| pair | co invoices | all sale invoices | support | confidence a to b | lift |
|---|---|---|---|---|---|
| 20728 + 22384 | 502 | 19,960 | 0.02515 | 0.4365 | 7.994 |
| 20727 + 22384 | 544 | 19,960 | 0.02725 | 0.4273 | 7.825 |
| 21931 + 22411 | 524 | 19,960 | 0.02625 | 0.4426 | 7.518 |
| 23203 + 23209 | 504 | 19,960 | 0.02525 | 0.4094 | 7.443 |
| 20728 + 22383 | 539 | 19,960 | 0.027 | 0.4687 | 7.286 |
| 20728 + 22382 | 483 | 19,960 | 0.0242 | 0.42 | 7.246 |
| 22960 + 22961 | 472 | 19,960 | 0.02365 | 0.417 | 7.162 |
| 22382 + 22383 | 533 | 19,960 | 0.0267 | 0.4607 | 7.161 |
| 20727 + 22383 | 584 | 19,960 | 0.02926 | 0.4588 | 7.131 |
| 21931 + 22386 | 515 | 19,960 | 0.0258 | 0.435 | 7.128 |
| 20727 + 20728 | 522 | 19,960 | 0.02615 | 0.4101 | 7.117 |
| 20725 + 22384 | 606 | 19,960 | 0.03036 | 0.3872 | 7.091 |
THE METHOD
From source to answer
Invoice presence matrix and pair support, confidence and lift among the 30 most frequent codes.
The Python source exposes this study’s transformations. The complete project download includes shared preparation and evaluation routines.
THE NEXT DECISION
What follows from the finding
Use a small merchandising experiment to test a selected pair; co-purchase alone does not establish an uplift.
Where the conclusion stops
Historical invoice lines; credits are not reliably matched to original sales. Gross purchases are not profit. Unidentified customers cannot support customer-level conclusions. Exact repeated lines remain unless the study explicitly compares removal.
Related studies may reuse observations or holdouts. These are historical analyses; associations and backtests do not demonstrate commercial impact. Further model tuning needs new, untouched evaluation data.