The Diagnostic Feature Designer app lets you work with ensemble data interactively and experiment with various processing and feature options. Once you have determined which features work best, you can generate code that reproduces your computations. This code allows you to apply the same computations to new or expanded ensemble data. You can use this code directly, or modify the code to suit your application. For more information on code generation in the app, see Automatic Feature Extraction Using Generated MATLAB Code.
|Diagnostic Feature Designer||Interactively extract, visualize, and rank features from measured or simulated data for machine diagnostics and prognostics|
|Manage ensemble data stored in the MATLAB workspace using code generated by Diagnostic Feature Designer|
|Find the workspace ensemble member indices for members that match a specified variable name and value|
|Return ensemble member data based on the member index|
|Update a workspace ensemble with partitions of modified or added data computed in parallel processing|
|Write data to a specific workspace ensemble member|
|Read all data in datastore|
|Read feature values, independent variables, and condition variables from an ensemble data set into a table|
|Extract data from an ensemble member given a path|
|Reset datastore to initial state|
|Unique values in array|
|Write data to member of an ensemble datastore|
|One-way analysis of variance|
|One-dimensional Bhattacharyya distance between two independent data groups to measure class separability|
|Receiver operating characteristic (ROC) curve or other performance curve for classifier output|
|Wilcoxon rank sum test|
|One-dimensional Kullback-Leibler divergence of two independent data groups to measure class separability|
|Adjust feature ranking scores using correlation factor|
Diagnostic Feature Designer can generate code that reproduces your interactive computations and allows you to automate feature extraction on similar input data. You select among your features, computed variables, and ranking tables to specify what the code includes.
This example shows how to extract features from measurement data in the app and generate a MATLAB function that reproduces the calculations for those features.
This example shows how use a reduced data set to develop features in Diagnostic Feature Designer, generate code to automate the feature computations you select, and then use that code with the full-member data set to compute a feature set for model training.
Explore the functional organization of MATLAB code generated in Diagnostic Feature Designer.
Algorithm design with Predictive Maintenance Toolbox™ uses data organized in ensembles. You can generate ensemble data from a Simulink® model or create ensembles from existing data stored on disk.