|Friday, February 21|
|CS16 Implementing the Tools||
Fri, Feb 21, 3:15 PM - 4:45 PM
Predictive SPC (302696)Josep Ferrandiz, MetaScale
*Alex Gilgur, Google
Keywords: statistical process control, nonparametric, model, prediction, regression, quantile, nonlinear, Six Sigma, distribution, non-gaussian, alerting, control
Statistical process control (SPC) is a well-described framework used to identify weak points in any process and predict the probability of failure in it. The distribution parameters of process metrics have been translated into process capability, which evolved in the 1990s into the Six Sigma methodology in a number of incarnations. However, all techniques derived for SPC have two important weaknesses: They assume the process metric is in a steady state and they assume the process metric is normally distributed, or can be converted to a normal distribution. These shortcomings threaten to make it irrelevant to any fast-changing environment (IT, economy, advertising, etc.). This paper presents an innovative way to overcome them and demonstrates how we have implemented it in an IT production environment.