Log Data Extraction Using Machine Learning for Advertising Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for optimizing advertising placement in digital marketing face difficulties in efficiently analyzing log data due to the enormous number of combinations of item values, making it challenging to improve planning and implementation efficiency.
Innovation Solution
A computer-readable recording medium stores an extraction program that generates combinations of conditions from item values, calculates an index value using a machine learning model, and extracts specific combinations based on the index value, enabling effective evaluation of hypothesis significance and improving planning efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all kinds of combinations of item values are analyzed by logistic regression, then the significance of each item value can be calculated, but the enormous number of combinations makes it difficult to perform the analysis
Solution Approach 1:
The patent segments the analysis process into two stages: first using association rule mining to identify promising combinations, then applying logistic regression only to those selected combinations. This segmentation reduces the number of combinations requiring computationally intensive logistic regression analysis while maintaining significance calculation accuracy for the most relevant combinations.
Solution Approach 2:
The patent performs preliminary filtering using association rule mining before conducting logistic regression analysis. By pre-selecting combinations that meet certain criteria (such as support and confidence thresholds), the system reduces the search space and prepares a refined set of combinations for subsequent significance analysis, improving overall analysis efficiency.
2Productivity
If the number of combinations of item values is reduced, then the analysis can be performed more efficiently, but it becomes difficult to evaluate all possible hypotheses
Solution Approach 1:
The analysis is divided into two phases: an exploratory phase using association rule mining to identify promising combinations across all item values, and a detailed evaluation phase using logistic regression on selected combinations. This segmentation allows comprehensive hypothesis generation while maintaining efficient evaluation of the most promising hypotheses.
Solution Approach 2:
Association rule mining acts as an intermediary between the full set of item value combinations and the logistic regression analysis. It filters and ranks combinations based on preliminary metrics (support, confidence), creating a bridge that enables efficient processing while maintaining coverage of important hypotheses through the two-stage approach.
Data Source
AI summary
A non-transitory computer-readable recording medium stores therein an extraction program that causes a computer to execute a process including: generating a plurality of combinations of conditions relating to a plurality of item values included in data; calculating an index value that indicates a degree of cooccurrence between a specified response variable and each of the plurality of combinations, by using a machine learning model that estimates a response variable from the plurality of item values, the machine learning model having been trained by using the data; and extracting a specific combination from among the plurality of combinations based on any one of the condition and the index value.