Insights: Software that extracts knowledge from data. Model. Predict. Simulate. Discover. Associate. Gain new Knowledge and Insights from noisy data. Easily and Reliably. Ultra-fast, parallel, self-organizing, high-dimensional modeling of real-world processes and complex systems. INSIGHTS
Online Store.
Test your copy.

Self-configurable Editions.
Insights Free to Insights Ultimate
Powerful entry-level knowledge discovery of your data to professional multi-step ex ante forecasting of interdependent systems and high-performance, high-dimensional modeling for identification, classification and forecasting problems of complex systems.
Download Copy Now

Features that all Editions Share.
Self-organizing Modeling
builds explicit and complexity minimized mathematical models, self-selects inputs and displays their importance
F, I, C, EQU (*)
64-bit Parallel Software
out-of-the-box multi-core support that optimally scales to your machine
Composite models
provides per-sample prediction interval
High-dimensional modeling
auto-detecting, no explicit feature selection required
Multi-level model validation
optimal noise filtering to avoid overfitting
Live prediction validation
checks model stability and reliability of predictions
Excel support (optional)
for data import and model export (v2016 or newer; Excel is not required to use all other features of Insights)
(*) F: Forecasting; I: Identification; C: Classification; EQU: explicit analytical model equation that describes the data

Configurable Features.
Similar Patterns
forecast complex time processes without models
Autoupdate Forecasts
automatically run simulations or forecast real-time data stored in Excel on Similar Patterns models in Insights
up to 120 steps ahead
are stored
Cost-sensitive Modeling
build tailored, purpose-driven, prevalence-sensitive classification models from imbalanced data sets including ROC and Cost Curve plots
Systems of Equations including VAR models
self-organizing modeling of interdependent complex systems for automated ex ante (scenario) forecasting
Model Export
ready-to-use models in different formats for integration in web apps, workflows or further external analysis and use
Excel, Python, MATLAB/Text, Objective-C, AppleScript, Systems of equations in Ecxel
Number of potential inputs supported
for direct use in model self-organization, including time lags for dynamic models. Key for model accuracy.
20 000
Number of samples supported
for model self-organization. Build reliable models from small samples sizes from noisy data (very well on >1000, already).
1 000 000
Number of forecasting steps supported
including live prediction validation

Download Copy Now
Deep Learning and noise immunity of GMDH adaptive self-learning and self-organizing data mining, forecasting and classification of streams of data