Online Recommendation System Using Time-Series Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In online retail environments, users are often presented with inaccurate item recommendations that do not account for purchases made in physical stores, quantity, or time since purchase, leading to irrelevant suggestions.
Innovation Solution
A method and system that generate time-series models using ARIMA to predict future purchases by analyzing past sales data, incorporating survival analysis to determine repurchase probability, and creating an ensemble model for recommending items based on historical purchase patterns, ensuring relevant and timely suggestions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If items previously purchased are re-presented to the user without regarding to quantity or time since purchase, then the system is simple to implement, but the recommendation accuracy deteriorates
Solution Approach 1:
The patent applies parameter changes by incorporating time-based parameters (time since last purchase, frequency of purchase) and quantity parameters into the recommendation system. The system modifies the presentation of items based on these changing parameters, such as adjusting recommendation priority based on how long it has been since the item was purchased or how much was bought previously, thereby improving accuracy while maintaining manageable complexity through structured parameter handling
2Ease of manufacture
If only items purchased online are considered for recommendations, then the system is easy to implement, but the recommendation accuracy deteriorates due to exclusion of physical store purchases
Solution Approach 1:
The patent merges data from multiple sources including online purchases and physical store purchases into a unified recommendation system. By combining these data sources, the system gains a more complete view of customer purchasing behavior, improving recommendation accuracy without significantly increasing implementation complexity through integrated data processing
3Measurement precision
If a comprehensive time-series model with multiple features is generated, then the recommendation accuracy is improved, but the system complexity increases
Solution Approach 1:
The patent segments the recommendation system into distinct modular components: data collection modules (online and offline purchase tracking), time-series analysis modules (ARIMA modeling), feature extraction modules, and recommendation generation modules. This segmentation allows each component to be developed, tested, and optimized independently, managing overall system complexity while achieving high accuracy through coordinated operation of specialized modules
Data Source
AI summary
Methods and systems for predicting relevant items to be presented to a user in an online environment are described. The methods and systems described herein generate models based on previous item selections to determine an overall time series model for predicting a relevant time of next item selection as well as items most likely to be selected at that time. Complementary items can be presented to the user alongside the selection of most relevant items.