Validate Preprocessing

In pguIMP: 'pguIMP'

Validate Preprocessing Help

Uni- and bivariate analyses validating the results of the preprocessing procedure.

Menu

Feature

Selects a feature for univariate graphical analysis.

Validate

Performs the validation analysis.

Analysis

Validation analysis results

Graphical validation of substitution of imputation sites. Left: Pareto density function of the selected feature's instance value distribution. Middle: Boxplot of the selected feature's instance value distribution. Upper right: Cumulative density function of the selected feature's instance value distribution. Lower Right: qq-plot of the selected feature's instance value distribution. The original distribution and the distribution with substituted imputation sites are color coded.

Test results

Univariate comparison of the distributions of the respective feature instances in the original condition and with substituted imputation sites. The distributions are compared using the Kolgomorov-Smirnov test and Wilcoxon's rank test.

Central moments

Table showing the central moments of the instance value distribution of each feature for either the original data or the date with substituted imputation sites. The deviation table shows the absolute deviation of the central moments of both data sets.

Correlation Validation Analysis

Bivariate cross-wise correlation analysis between the individual features of a data set. The plot shows the correlation R-values between the individual features. Plotted are the R-values of the dataset with substituted imputation sites against those of the original dataset. The box plot below shows the distribution of the difference between the r-Values obtained from both data sets. Next to the Boxplot the statistics of the distribution shown in the box plot are listed. The data underlying the correlation validation analysis is shown in the Correlation Validation Data table.

pguIMP documentation built on Sept. 30, 2021, 5:08 p.m.