Usage Prediction Method for Resource Time-Series Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing prediction models for resource usage in information processing systems struggle to accurately reflect changes in usage trends specific to time zones and days of the week, leading to reduced prediction accuracy.
Innovation Solution
The method involves classifying time-series data into groups based on attributes like day of the week and time zone, generating separate time-series data for each attribute, and creating predicted value calculation expressions using time-series analysis or machine learning for each group, allowing for more accurate predictions by reflecting specific trends.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single prediction model is used for all time periods, then the model is simple and easy to maintain, but it cannot accurately reflect usage trends specific to different time zones and days of the week
Solution Approach 1:
The patent divides the time-series data into multiple groups based on time attributes (day of the week, time zone). Separate prediction models are generated for each group, allowing each model to capture the specific usage patterns of its corresponding time period. This segmentation resolves the contradiction by trading increased model complexity for significantly improved prediction accuracy for each specific time period.
Solution Approach 2:
The patent applies different prediction models to different time periods based on their specific characteristics. Each local time period (e.g., weekdays in Asia, weekends in America) receives a customized model that reflects its unique usage patterns, rather than applying a single uniform model. This local quality approach enables accurate reflection of regional and temporal usage trends.
2Measurement precision
If separate prediction models are generated for each time attribute group, then prediction accuracy for specific trends improves, but the complexity of data processing and model generation increases
Solution Approach 1:
The patent performs preliminary classification of time-series data into groups based on time attributes before generating prediction models. By pre-organizing the data according to day of the week and time zone, the system simplifies the subsequent model generation process. This preliminary action makes the complex task of generating multiple models more manageable and systematic.
Solution Approach 2:
The patent uses a universal data processing framework that can handle multiple time attributes (day of the week, time zone) in a consistent manner. The same prediction algorithm is applied across different groups, with the universality of the framework reducing the overall complexity despite the multiplicity of models.
3Measurement precision
If time-series data is classified and processed separately for each attribute, then the reflection of change trends specific to each group improves, but the processing time and computational resources increase
Solution Approach 1:
By segmenting the time-series data into distinct groups based on time attributes, the patent enables parallel processing of different data groups. Each segment can be processed independently and simultaneously, reducing the overall processing time compared to sequential processing of the entire dataset. This segmentation strategy resolves the contradiction between detailed trend analysis and processing efficiency.
Data Source
AI summary
A usage prediction method executed by a computer, the usage prediction method includes classifying a plurality of records corresponding to a plurality of times included in first time-series data indicating a history of usages of a resource into a plurality of groups respectively corresponding to attributes of the plurality of times; generating second time-series data for each attribute by combining the records belonging to the group corresponding to the same attribute for the plurality of classified groups in order of the times; generating, for each attribute, an expression for calculating a predicted value to be used for calculating a predicted value of the usage based on the generated second time-series data; and calculating the predicted value of the usage based on the expression for calculating the predicted value for each attribute.