Time-Factored Performance Prediction via ML
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing performance prediction techniques lack reliability and scalability due to subjective factor selection and labor-intensive data collection, leading to varying conclusions among analysts and constrained data usage across domains.
Innovation Solution
A machine learning model is trained using query intents and performance results from a query-URL click graph, allocated into time intervals based on milestones, to predict target entity performance by inputting target query intents into the model before a target milestone, thereby reducing subjectivity and increasing data scalability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If analysts use subjective factor selection and manual data collection methods, then they can perform performance prediction analysis, but the reliability and scalability of the prediction system deteriorates due to varying analyst judgments and labor-intensive processes
Solution Approach 1:
The patent replaces the mechanical system of manual analyst judgment and data collection with an automated machine learning model. The system automatically extracts performance-influencing factors from unstructured data sources, allocates them to time intervals based on milestones, and generates predictions without human intervention, thereby improving reliability while maintaining low automation in the original approach
Solution Approach 2:
The machine learning model performs self-service by automatically selecting and weighting performance factors, collecting and processing data from multiple sources, and generating predictions without requiring analyst input. This automated self-service approach eliminates subjectivity and improves prediction reliability consistently across different cases
2Productivity
If analysts manually identify and analyze performance-related data, then they can perform detailed analysis, but the productivity and scalability of the system deteriorates due to labor-intensive processes
Solution Approach 1:
The system performs preliminary action by pre-processing and organizing performance data into time intervals and milestones before prediction is needed. Training data is pre-allocated to time intervals, and the machine learning model is trained in advance, so that when prediction is required, the system can quickly generate results without manual data collection and analysis time
Solution Approach 2:
The patent replaces the manual mechanical process of data collection and analysis with automated computer-based processing. The system automatically queries multiple data sources, extracts relevant performance factors, allocates them to time intervals, and processes them through the machine learning model, dramatically improving productivity while reducing the time loss associated with manual analyst work
3Measurement precision
If analysts use subjective factor importance assignment, then they can adapt to different cases, but the measurement precision and consistency of predictions deteriorates due to varying analyst judgments
Solution Approach 1:
The system uses parameter changes by dynamically adjusting the importance weights of different performance factors based on the specific case and domain. The machine learning model automatically determines optimal weights for each factor during training, allowing precise measurement adapted to different cases without subjective analyst judgment. This resolves the contradiction by making precision consistent while maintaining adaptability through data-driven parameter adjustment
Data Source
AI summary
Training query intents are allocated for multiple training entities into training time intervals in a time series based on a corresponding query intent time for each training query intent. Training performance results for the multiple training entities are allocated into the training time intervals in the time series based on a corresponding performance time of each training performance result. A machine learning model for a training milestone of the time series is trained based on the training query intents allocated to a training time interval prior to the training milestone and the training performance results allocated to a training time interval after the training milestone. Target performance for the target entity for an interval after a target milestone in the time series is predicted by inputting to the trained machine learning model target query intents allocated to the target entity in a target time interval before the target milestone.