Methodology for identifying emerging issue based on fused severity and sensitivity of temporal trends

Abstract

The invention relates generally to a method for identifying emerging issue based on fused severity and sensitivity of temporal trends, specifically a method for temporal trend detection employing non-parametric techniques. A set of discrete data is provided and a rank is assigned to the data based on both sensitivity and severity of the data. The method statistically ranks the ranked data by categorizing the data in bins defined by an average positional ranking that identifies the severity of the data for each sensitivity category provided by a bin. The method then clusters the statistically ranked data that has been categorized by average positional ranking so as to detect changes in the data. Clustering the statistically ranked data can include using a multi-nominal hypothesis testing procedure. The method then identifies trends in the data based on the detected changes.

Description

technical field [0001] The present invention relates generally to methods for temporal trend detection using nonparametric techniques, and more particularly to using the sensitivity and significance of data to extract temporal trends and classify the trends in various ways by employing nonparametric techniques to allow for different data-driven decision-making methods. Background technique [0002] The collection of product or process data and its analysis enables users to make various data-driven decisions. Examples include warranty and repair data collected by product companies, demographic data collected by states, and weather data collected by weather scientists. The purpose of collecting and interpreting this product or process data is to reduce costs (both tangible and intangible) through early detection of emerging issues. Due to the nature of the data itself, data collection limitations, or data storage limitations, the collected data is often of a discrete nature,...

AI Technical Summary

Problems solved by technology

Additional limitations of known trend detection methods include: (1) the methods do not incorporate the sensitivities and significance of variables to detect and classify trends; (2) the methods generally assume that the data come from a parametric distribution, which sometimes may not be correct assumptions; (3) the method does not perform within-cluster analysis to provide causal (physics-based) and non-causal relationships for variables within each cluster; (4) the method uses thresholds to classify trends classification, thus needing to develop sufficient confidence levels to balance Type 1 / Type 2 errors; and (5) any missing data are interpolated, leading to inaccuracies associated with the interpolated values

Images