Time-Aware Feature Engineering to Prevent Target Leakage

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing machine learning solutions require significant human intervention and domain knowledge for feature engineering, leading to labor-intensive and error-prone processes that are not scalable and prone to target leakage.

Innovation Solution

A model training and deployment system that automatically generates and detects features from multiple datasets, minimizing human input and reducing errors by using time-awareness to avoid target leakage, while scaling with dataset volume and variety.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual feature engineering is used to combine data from multiple datasets, then domain knowledge and human judgement can be applied to create meaningful features, but the process becomes labor-intensive and error-prone

Engineering Contradiction:
Improveaccuracy of feature engineeringVSAvoidefficiency of feature engineering
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system performs self-service by automatically detecting relationships between data sources and generating features without requiring manual human intervention. The automated feature engineering system identifies how different data sources are related and combines them to create training datasets, eliminating the labor-intensive manual process while maintaining consistency and reducing errors through systematic automated detection and generation.

Inventive Principle:
Principle #25Self-service

2Productivity

If automated feature engineering is implemented to reduce manual labor, then productivity and consistency improve, but the system becomes more complex and requires sophisticated algorithms

Engineering Contradiction:
Improveefficiency of data science solutionVSAvoidcomplexity of automated system
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system uses an intermediary approach by introducing an automated feature engineering system that acts as a mediator between raw data sources and machine learning models. This intermediary automatically detects relationships between data sources, combines data, and generates features, simplifying the overall process while managing complexity through systematic automated detection and generation algorithms.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Adaptability or versatility

If data from multiple datasets is combined manually to create training datasets, then flexibility in handling different data types is maintained, but the process is time-consuming and scales poorly

Engineering Contradiction:
Improveability to handle diverse data typesVSAvoidtime required for data combination
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system performs self-service by automatically detecting relationships between data sources and generating features without requiring manual human intervention. The automated feature engineering system identifies how different data sources are related and combines them to create training datasets, eliminating the labor-intensive manual process while maintaining consistency and reducing errors through systematic automated detection and generation.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system performs preliminary action by pre-processing and automatically combining data from multiple sources before model training. The automated feature engineering detects relationships and generates features in advance, creating ready-to-use training datasets that can be directly fed into machine learning models, thereby eliminating time-consuming manual data combination steps.

Inventive Principle:
Principle #10Preliminary action

4Reliability

If manual feature engineering is performed to ensure data quality, then errors like target leakage can be avoided through human judgement, but the process requires significant human resources

Engineering Contradiction:
Improvequality of training datasetVSAvoidhuman resources required
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system performs self-service by automatically detecting relationships between data sources and generating features without requiring manual human intervention. The automated feature engineering system identifies how different data sources are related and combines them to create training datasets, eliminating the labor-intensive manual process while maintaining consistency and reducing errors through systematic automated detection and generation.

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS12632778B2Automated feature engineering for machine learning models
Publication Date: 2026.05.19 DATAROBOT INC
  • US12632778B2 patent drawing
  • US12632778B2 patent drawing
  • US12632778B2 patent drawing

AI summary

Training computer models by generating time-aware training datasets is provided. A system receives a secondary dataset to be combined with a primary dataset for generation of a training dataset. The primary dataset includes a plurality of data records where at least one data record corresponds to a time-of-prediction value corresponding to a timestamp at which at least one data record was used to generate a prediction. The secondary dataset includes a plurality of features where at least one feature corresponds to a timestamp value. The system selects a feature within the secondary dataset with a timestamp that precedes or matches a time-of-prediction value for a corresponding data record within the primary dataset. The system generates the training dataset that includes the primary dataset and the selected feature. The system trains a model using the generated training dataset.