Quantitative Approach to Market Basket Analysis: Uncovering Patterns in Consumer Behavior
Laurila, Inari (2025)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025052013404
https://urn.fi/URN:NBN:fi:amk-2025052013404
Tiivistelmä
Web store purchase data can be used to analyze customer behavior and support the development of marketing strategies. In this study, a market basket analysis was conducted using purchase data from a Finnish web store. The objective was to identify product associations that could support marketing and sales efforts. A quantitative approach was applied, utilizing association rule mining techniques to detect patterns in purchasing behavior.
The analysis revealed that strong product associations could not be identified on a broad scale. While some products were frequently purchased together, their connections to other products were generally weak. Most product pairs showed little to no significant association (with lift values close to or below one), suggesting that these purchases were likely made by chance. As a result, developing a full-scale recommendation system was not considered appropriate.
Based on the findings, light development suggestions were made for the web store, such as displaying related products on product pages and considering time-specific promotional campaigns. Although the analysis did not uncover broad predictability, it did identify some useful product connections. Strategically leveraging this information may help improve customer experience, increase sales, and enhance the effectiveness of marketing efforts. Further research and experimentation with different actions are recommended based on the results.
The analysis revealed that strong product associations could not be identified on a broad scale. While some products were frequently purchased together, their connections to other products were generally weak. Most product pairs showed little to no significant association (with lift values close to or below one), suggesting that these purchases were likely made by chance. As a result, developing a full-scale recommendation system was not considered appropriate.
Based on the findings, light development suggestions were made for the web store, such as displaying related products on product pages and considering time-specific promotional campaigns. Although the analysis did not uncover broad predictability, it did identify some useful product connections. Strategically leveraging this information may help improve customer experience, increase sales, and enhance the effectiveness of marketing efforts. Further research and experimentation with different actions are recommended based on the results.