GMDH Streamline

Forecasting software can be employed in a wide variety of branches of human activity, from high-tech biological analyses to simple games. Although the results depend largely on the underlying input assumptions, a sharp increase in the use of these tools has been witnessed in recent times and GMDH Streamline (formerly GMDH Shell) is a practical prediction program that seeks to capitalize upon this trend.

The application features two modules, one geared towards the business environment while the other is better suited for scientific statistical analyses. The “Business Forecasting” module is best employed for sales or demand forecasting, while the “Data Science” component is more reliant on regression models and is thus best used for contingency analyses.

That being said, the two are just specific applications of a common framework, as the overall feel and power of the program is identical in both applications. As with any statistical program, the tool requires a source for all the input data and viable options are simple CSV, TXT or Excel documents, as well as ODBC or OLEDB connections.

The program automatically extracts all the valid information from source containers and a “Data Explorer” tab allows users to check their input information. This step should not be taken lightly, as the performance of the output model is contingent upon the soundness of the source data.

A built-in wizard guides users through all the necessary steps, although one can also perform any adjustments deemed correct on the input variables. Various transformations can be applied to the target values, such as derivatives, averages or weights by time and creating predictions requires only that one clicks on the “Forecast” button.

Linear regression equations and data plots are automatically generated and users also have the option of saving the graphs to local PNG or vector EMF files.

In conclusion, GMDH Streamline is a comprehensive and practical tool for anyone in need of a forecasting application. The program allows its users to load datasets from local TXT or CSV documents, as well as from ODBC or OLEDB connections and create predictive models, complete with data plots.