Who Uses This Software?

Organizations that have a need for accurate, reliable forecasts and forecasting processes automation.

Average Ratings

2 Reviews

  • 4.5 / 5

  • 4 / 5
    Ease of Use

  • 4.5 / 5
    Customer Service

Product Details

  • Starting Price
  • Pricing Details
    Free Academic Licensing Program
  • Free Version
  • Free Trial
  • Deployment
    Installed - Windows
  • Support
    Business Hours

Vendor Details

  • GMDH
  • www.gmdhsoftware.com
  • Founded 2009
  • United States

About GMDH Shell

GMDH Shell is the easiest way to accurately forecast time series, create classifiers and regression models. Based on artificial neural networks, it allows you easily create predictive models, as well as preprocess data with dead simple point-and-click interface. Unlike other NN-based tools, it's very fast because of state-of-the-art parallel processing and great core algorithms optimization.

Spotlight media placeholder

GMDH Shell Features

  • Ad hoc Analysis
  • Ad hoc Query
  • Ad Hoc Reports
  • Benchmarking
  • Budgeting & Forecasting
  • Dashboard
  • Data Analysis
  • Data Visualization
  • Key Performance Indicators
  • OLAP
  • Performance Metrics
  • Predictive Analytics
  • Profitability Analysis
  • Strategic Planning
  • Trend / Problem Indicators
  • Data Extraction
  • Data Visualization
  • Fraud Detection
  • Linked Data Management
  • Machine Learning
  • Predictive Modeling
  • Semantic Search
  • Statistical Analysis
  • Text Mining
  • Analytics
  • Content Management
  • Dashboard Creation
  • Filtered Views
  • OLAP
  • Relational Display
  • Simulation Models
  • Visual Discovery
  • AI / Machine Learning
  • Benchmarking
  • Data Blending
  • Data Mining
  • Demand Forecasting
  • For Education
  • For Healthcare
  • Modeling & Simulation
  • Sentiment Analysis

GMDH Shell Reviews

Capterra loader

It represents a good option to quickly make simulations and useful forecasts for decision making.

Mar 15, 2018

4 / 5
Ease of Use

4 / 5
Features & Functionality

4 / 5
Customer Support

5 / 5
Value for Money
Likelihood to Recommend: 7.0/10 Not

Pros: The aspect that impresses me the most is its search scheme of the model that best fits the actual data that is available. Based on extremely simple models, which are evaluated to determine if the general error between the real and forecasted data is lower than the user defines as satisfactory, it evolves towards increasingly complex models (higher and non-linear orders) until achieving the desired precision.

Cons: When using the model to find the absolute maximum of a series of data where there are several scattered values throughout the calculation domain, with values very close to the absolute maximum, the model did not converge to the absolute maximum but to a local maximum, suffering from the same lack of other optimization methods.

In this regard I prefer to continue using other techniques or optimization methodologies such as Genetic Algorithms.

With regard to predictive series with stochastic components, when compared to methodologies such as "Box-Jenkins" it did not bring me substantial improvements.

Overall: Although for my practical purposes in the business environment the software does not represent a significant advance, it has been very practical for teaching purposes in the university because its interface is extremely easy to use.

In the business field where I have been useful is in the management, maintenance and forecasting of inventories.

Great software for rapid analysis and predictive prototyping

Jan 28, 2016

4 / 5
Ease of Use

5 / 5
Customer Support

Comments: GMDH Shell (GS) can take a raw somewhat messy data set in CSV and provide a predictive model more more quickly than just about anything and at a reasonable cost. The provided templates are a great start. There are quite a few options to adjust later to improve your models. And you can customize the templates to your preferences.

It is great how GS searches for the best model with the best algorithm for your problem. This is a feature normally reserved for the premium software tools.

You should have an understanding of inputs, targets, transformations, regression, forecasting, and classification modeling to get started. The available transformations are very useful especially the time series Lag. You will want to follow the documentation and tutorials to get the most out of it.

I recommend enabling the Experimenter's Layout and the Processing Results for Importance of Variables and Residuals in every default template.

With practice using GS is beneficial and fun.