Introduction

GMDH Shell (shortly GS) is a predictive modeling tool that produces mathematical models and predictions. You can start with data exploration and visualization, then design a new modeling experiment and finally automate your everyday data analysis and forecasting tasks.

When a new dataset is imported GS pops-up a list of preconfigured templates allowing you to choose between Regression, Classification and Time Series Forecasting tasks. The following examples should help identify your modeling task:

Please note that by default GS reads data series from columns. However horizontally aligned data can also be imported using the Transpose check-box in the Import dialog. Read more about how to prepare and import your dataset in the Import module documentation.

It is recommended to keep your data file in a separate directory where GS can store all project settings and modeling results. So a new GS project is just a separate folder where all data files and settings are located.

If you want to open this new project click at the Import button, select your data file located in the project directory and click OK. In the Import dialog check if GS reads your data correctly and if necessary adjust import settings in the dialog. Your selection of import settings will be saved to the project folder and applied to all files within the project folder.

If you configured the Import module to read from a new data source and project location then the next dialog will ask you to load one of the built-in templates, i.e. Regression, Classification and Time Series Forecasting.

When import module configuration and template selection are completed an important stage is to select Input variables and one or more predicted (Target) variables in the Data manager tab. Read more about Inputs, Targets and optional Transformations in the Preprocessor module documentation.

Then you can click at the Start button and obtain models and predictions.