Data-Driven Prediction Model Recoding via Semantic Hierarchies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data-driven prediction models face challenges in interpreting results due to numerous category levels of input and output variables, which complicates the optimization of prediction performance and does not effectively utilize semantic models for variable recoding.
Innovation Solution
A method that semantically annotates input data using hierarchical trees representing ontology elements, automatically recodes discrete values by cutting off hierarchical levels, and trains multiple machine learning models on modified data sets to determine the optimal prediction model with the highest quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual recoding of category levels is performed by data scientists, then prediction quality can be optimized, but time consumption and labor requirements increase
Solution Approach 1:
The system performs automated recoding of category levels using machine learning algorithms that independently analyze the data and determine optimal recoding strategies without requiring manual intervention from data scientists, thereby reducing time consumption while maintaining prediction quality
Solution Approach 2:
The system automatically modifies the parameter of category level recoding by transforming discrete values into recoded categories based on learned patterns from the data, enabling the system to adapt the data structure dynamically without manual recoding operations
2Loss of time
If automated approaches are used to optimize prediction models, then time consumption is reduced, but variable recoding capability is lost
Solution Approach 1:
The machine learning system automatically performs variable recoding as part of its data processing pipeline, enabling it to adapt the data structure dynamically without manual intervention while maintaining the benefits of automation
Solution Approach 2:
The system dynamically adjusts the recoding strategy based on the characteristics of the input data and the specific prediction task, allowing it to adapt variable recoding to different scenarios automatically rather than using fixed predefined recoding rules
3Reliability
If semantic models are integrated into the prediction process, then domain expertise is captured, but system complexity increases
Solution Approach 1:
The system uses semantic models as an intermediary layer that bridges raw data and machine learning algorithms, translating domain knowledge into a structured representation that guides the recoding process without requiring direct complex integration of domain expertise into the algorithms
Solution Approach 2:
The semantic models are used to pre-process and structure the data before it enters the machine learning pipeline, capturing domain expertise in advance through ontological frameworks that organize concepts and relationships, thereby simplifying the subsequent machine learning processing
Data Source
AI summary
Provided is a method for computer-implemented determination of a data-driven prediction model. The method processes digital input data having input and output variables and being semantically annotated based on a digital semantic representation having a hierarchical tree structure where each tree in the structure represents an input variable of the data, the leaf nodes of the respective tree being the discrete values of the input variable. The method of the embodiment provides a recoding of those discrete values by cutting off hierarchical levels of the respective trees. Based on this recoding, a plurality of data modifications is determined for the input data. Those data modifications are trained by a machine learning method where the trained machine learning method with the highest prediction quality is derived from the trained machine learning methods.