Outlier Processing for Device Performance Data

Solution Overview

Problem

Outlier data points in data sets can lead to erroneous conclusions when used to infer device operational characteristics, making it challenging to accurately represent device properties and trends.

Innovation Solution

A method to transform the original data set by identifying and removing outliers based on their deviation from a known trend, using weighted filtering and blame assignment algorithms to isolate points that do not conform to the trend, resulting in a more accurate representation of device performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If outlier data points are included in the data set, then the data set represents all observed data, but erroneous conclusions are drawn about device operational characteristics

Engineering Contradiction:

Improveaccuracy of device performance inferenceVSAvoidinformation loss from removing outliers

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

The patent extracts and removes outlier data points from the original data set through a systematic process: identifying points that deviate from the expected trend, calculating their impact on performance metrics, and selectively removing them to produce a cleaned data set that accurately represents device operational characteristics without being skewed by anomalous measurements

Inventive Principle:

Principle #2Taking out (Extraction)

2Measurement precision

If data processing is performed to remove outliers, then measurement precision is improved, but processing time and computational complexity increase

Engineering Contradiction:

Improveprecision of device characteristic measurementVSAvoidtime consumed by outlier removal process

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary actions by pre-defining trend expectations and thresholds before processing the data set. The system establishes what constitutes normal operational behavior in advance, enabling rapid identification and removal of outliers without requiring complex real-time analysis, thus improving measurement precision while controlling processing time

Inventive Principle:

Principle #10Preliminary action

3Productivity

If all data points are used for analysis, then the analysis is simple and quick, but the results are skewed by anomalous values

Engineering Contradiction:

Improvespeed of data analysisVSAvoidaccuracy of performance characterization

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The patent extracts and removes outlier data points from the original data set through a systematic process: identifying points that deviate from the expected trend, calculating their impact on performance metrics, and selectively removing them to produce a cleaned data set that accurately represents device operational characteristics without being skewed by anomalous measurements

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS8311772B2Outlier processing

Publication Date: 2012.11.13 TERADATA US INC

AI summary

Apparatus, systems, and methods may operate to acquire an original data set comprising a series of data points having an independent portion and a dependent portion, the dependent portion representing a measure of device performance that depends on at least one device characteristic represented by the independent portion. Additional activity may include identifying outlier data points in the series by determining, in comparison with all other members of the series, whether the outlier data points conform to a known trend of the series; transforming the original data set into a transformed data set by removing the outlier data points from the series; and publishing the transformed data set. Other apparatus, systems, and methods are disclosed.