From Basket to Recommendation: Association Rule Learning with Python

2026-07-12 · Source: Machine Learning on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, long

Summary

This study details the development of a service recommendation system using Python and Association Rule Learning, applied to user and service data from the Armut platform. The process begins by uniquely identifying services by combining "ServiceId" and "CategoryId", then defining user purchases within the same month as distinct "baskets." The mlxtend library is utilized, with pandas for data manipulation. The Apriori algorithm identifies frequent service combinations with a "min_support" of 0.01. Association rules are then generated, and their strength is evaluated using Support, Confidence, and Lift metrics, with rules filtered for "support > 0.01", "confidence > 0.1", and "lift > 1". A Python function, "arl_recommender", is implemented to suggest up to three services based on a user's last purchased service, demonstrating an end-to-end recommendation workflow.

Key takeaway

For Data Scientists or Machine Learning Engineers building recommendation systems, you should consider Association Rule Learning for its ability to derive meaningful insights from co-occurrence patterns without complex models. Define "baskets" carefully based on user behavior, like monthly purchases, and combine "ServiceId" and "CategoryId" for unique item identification. Evaluate rules holistically using Support, Confidence, and Lift, not just Lift, to ensure practical relevance and avoid recommending obscure items.

Key insights

Association Rule Learning can build effective service recommendation systems by analyzing co-purchased items.

Principles

• Data representation is crucial for algorithm input.
• Evaluate rules using Support, Confidence, and Lift.
• User behavior drives recommendation systems.

Method

Combine "UserId" and "CreateDate" to define monthly "baskets," then use Apriori to find frequent itemsets and "association_rules" to generate recommendations based on Support, Confidence, and Lift.

In practice

• Combine "ServiceId" and "CategoryId" for unique service IDs.
• Filter rules by "support > 0.01", "confidence > 0.1", "lift > 1".
• Sort recommendations by Lift for stronger relationships.

Topics

• Association Rule Learning
• Apriori Algorithm
• Service Recommendation
• Python
• mlxtend
• Market Basket Analysis

Best for: Machine Learning Engineer, Data Scientist, AI Student

Related on AIssential

• Bi-NAS: Towards Effective and Personalized Explanation for Recommender Systems via Bi-Level Neural Architecture Search — The Bi-NAS framework optimizes recommender system explanations by combining bi-level neural architecture search with LLM-enhanced, personalized justification generation.
• Athletes’ performance and injury management in sports training using association rules and data mining techniques — Association rule mining reveals interpretable patterns in sports data, linking training factors to performance and injury outcomes.
• Machine Learning for Beginners: Start Small, Learn Fast, Build Real Things — Machine learning teaches computers to find data patterns for decision-making, best approached by starting small with Python.
• CLARK: Closed-loop Learning for Adaptive Reasoning over Knowledge Graphs — CLARK integrates KGs, symbolic rules, and probabilistic reasoning for adaptive, interpretable, knowledge-driven classification.
• Building Pookie-Picks: A Memory-Efficient Content-Based Movie Recommender — Memory-efficient content-based recommenders can achieve fast lookups by combining NLP vectorization with memory-mapped similarity matrices.
• Public Machine Learning Solver Framework for Novices in the Machine Learning Domain — A new public framework guides ML novices through problem-solving by recommending complete, expert-informed pipelines.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning on Medium.