Time series anomaly detection, anomaly classification, and transition analysis using k-neighbors and logistic regression approach

摘要

A method and system for time series transition analysis of data is disclosed herein. The method is randomized using receiving time series data, generating a training data set that includes randomized data points, and using a set of randomized data points within a time window. Generating a combination of data points; calculating a distance value based on a randomized combination of data points; generating a classifier based on a plurality of calculated distance values; Using to determine the probability that new time series data generated during a new execution of the process matches the time series data. A system for performing this method is also disclosed. [Selection] Figure 2