When used in non-English locales, the Scoring Wizard could produce incorrect predicted values for Multilayer Perceptron Neural Network models. When the data being analyzed contain no system missing values, Codebook now does not report counts for system missing values.
That has been fixed.Ī pushed installation on Windows sometimes failed to launch. When Collaboration and Deployment Services connected to some Statistics Servers, the connection sometimes could not be made. Requests for a Non-Parametric > Related Samples test using the McNemar's test now produces the correct frequency counts in the Model Viewer.
On Mac OS, redundant Network License Managers could fail after a few minutes. On Mac OS, applying a template from Chart Builder or to Legacy graphics could cause the application to close suddenly. Requests for chi-square statistics in Crosstabs for a table larger than 2x2 now gives correct Exact significance on all platforms.
0 kommentar(er)
