Data Analysis#
Hypothesis testing and confidence intervals#
- Section
Presents tools that are used to infer underlying characteristics from sampled measurements. The tools include confidence intervals and hypothesis tests, and also sample-size calculations. Both the characteristics of a single sample and also comparisons between samples are covered.
There is overlap between that applications of these routines and ANOVA.
As a guide, the inference module handles statements about single populations,
comparisons between two populations, and simple non-parametric comparisons between multiple populations.
More detailed structues like data from factorial experiments, split-plots designs, etc.
are handled with ANOVA and regression techniques.
Regression and ANOVA#
- Section
Describes how to create linear models that explain the effect of many factors on a measurable response, including factors that are quantitative/continuous and also factors that are categorical (eg. one tool vs. another). MQR does not perform regression itself, but presents results in a consistent way. There are several tools in python that do regression; examples here are shown with statsmodels, and some tools in MQR rely on the output from statsmodels. This section also describes the analysis of how well a model explains the observations from which it was created. The mathematical tools are ordinary least squares (OLS) and analysis of variance (ANOVA). ANOVA can be viewed as a particular setup of an OLS problem. This section shows how to organise data ready for regression, and also describes the tools that MQR provides to analyse regression results.
Measurement system analysis#
- Section
Describes how to quantify the contribution of multiple sources of variability that contribute to uncertain measurements. The goal of measurement system analysis is to verify that a gauge and measurement process is sufficiently precise to (a) observe the changes that have to be made to improve a product or process and (b) keep both stable.