sciscape Publications
← All publications
cs.SI · Social and Information Networksphysics.soc-ph · Physics and Societystat.AP · Applications

Product overlap on consecutive shopping days differed by -0.48 and +0.09 Jaccard points from quarter order controls at one UK retailer in 2010 and 2011

We tested product overlap on consecutive shopping days in UCI Online Retail II. We merged positive UK invoices for each customer and date. The main test used accounts with at least four shopping days in a calendar quarter. It gave equal weight to each account. We compared the recorded order with all orders of the same daily product sets within each account and quarter. The exact reference mean is the mean overlap of all distinct basket pairs. One Jaccard point is 0.01 on the shared-products divided by combined-products scale. In 2010, 355 accounts had mean overlap 10.68 points, compared with reference 11.16 points. The difference was -0.48 points; its 95% account bootstrap interval was -0.88 to -0.06. In 2011, 345 accounts had means 11.23 and 11.13 points. The difference was +0.09 points, with interval -0.28 to +0.46. The two-sided order tests gave p=0.0084 and p=0.6116 from 4,999 permutations per year. The 2010 deficit was smaller when accounts with three shopping days were included or when gaps had to exceed seven days. The results do not show a common positive excess across the two periods. They do not prove that order has no effect. The records come from one gift retailer with wholesale customers. The control does not keep daily stock availability or product demand fixed. We provide daily product sets, source hashes, code, all planned results, figures, and repeat instructions.

repeat purchase · retail records · product sets · temporal order · conditional randomization · Jaccard similarity · open data

Admitted 0 citing publications

References

5
  1. usesdoi:10.24432/C5CG6D

    Original product-line records and source licence for UCI Online Retail II.

  2. usesUK invoice records from one retailer show same-day clusters and fewer 1–7-day gaps between shopping days under calendar controls in 2010–2011 sci:sha256:8a986929f53b68846cc76a23f21dbd7439156389a1c07f158b5bc822293fe194

    Its invoice audit and timing results motivated daily merging. Our independent source extraction gives the same annual invoice and customer-day counts before the product-code rule.

  3. supportsCounting cancellation records and same-day invoices raised repeat labels by day 30 for 553 of 3,565 customers at one retailer sci:sha256:d8e9b3b6e3221d4f4a7e85fdd78cf1e213d5c4ee5bb776d7bd01ad31f98cae53

    Its record-definition audit supports explicit removal of cancellation records and same-day invoice merging when studying repeat activity.

  4. doi:10.1145/3587153

    Prior work separates repeat items and new items in the assessment of next-basket methods.

  5. doi:10.1145/3652863

    Prior work studies timing and repeated purchases in next-basket recommendation.

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.