Performance Data Analytics

Performance Data Analytics

Attribute and Ordinal Data: 

Algorithms:

Probability of Success (PS) - a PS is calculated from a confidence level, number of samples, and number of successes.

Kolmogorov-Conover – this algorithm determines whether two sets of 1D ordinal data (ranked, numerical data) belong to the same population. This applies to pass-fail data as well.  

Implementations:

KC Download (compiled C++) – this console program implements the KC algorithm, and additionally calculates the probability of success (PS) for achieving each rank of the data. *See Readme within archive for more details.  

KC (LabView) – this program implements the KC algorithm with an associated LabView GUI.  

Continuous Data:

Algorithms:

Autocorrelation – an autocorrelation coefficient is calculated to quantify the “randomness” or independence of data samples.

Kolmogorov-Smirnov – this algorithm determines whether two sets of 1D continuous data belong to the same population.

Levene’s test – a modified version of this algorithm uses the Brown-Forsythe statistic for a more robust comparison of the sample variances of two datasets.

T-test – two versions of this test for comparing sample means with either equal or unequal sample variances.

Implementations:

KS Download (compiled C++) – this console program implements the aforementioned algorithms for continuous data, and the results are comprehensively reported in the console window. *See Readme within archive for more details.

Selected Publications: 

• Simplified Framework for Robot Coordinate Registration for Manufacturing Applications. Accepted for publication at IEEE International Symposium on Assembly and Manufacturing. 
• Computational Method to Apply Kolmogorov-Conover to Robot Performance Metric Evaluation. Publication pending.
• "Pass-Fail" Performance Measurement of Robot Pick-and-Place Applications. Publication pending. 
• Performance Methods and Comparative Analytics for Robot Pick-and-Place Operations. Publication pending.