AK Miner and Visual Pattern Explorer
Unearth the hidden relations in multivariate data. Identify the features that drive your objectives. Uncover their causal relations. As a result, make safer predictions.
Identify the Most Important Features
- Rank data features in order of importance using the AK Feature Importance Analyzer
- Rank these features globally (across all data) and locally (for a specific group of similar data items)
- Given these rankings and your specific goals pick one or more target features to discover the data feature configurations that offer the greatest predictive power
- Precisely discern the value ranges of these feature configurations using the AK Visual Pattern Explorer Dashboard
Find Patterns and Groups That Matter
-
The AK Pattern Discovery Engine can extract groups of similar data itemsMultivariate and high-dimensional data are ecosystems of subpopulations with different characteristics. Finding these groups of data items will allow you to specify dedicated strategies for each in order to maximize their effectiveness at economies of scale.that all have unusually high or low performance in a target feature of your choosing
- The AK Visual Pattern Explorer then guides you through the ecosystem of groups and patterns and allows you to form dedicated strategies and predictions
Ask What-If Questions
- Use the AK Causal Relation Extractor to combine the insights derived from the other tools in the AK Analyst
- Gain a more complete understanding on how the various features causally relate and interact
- Perform What-If Analyses to impose possible interventions and see how the target would change
- You may ask, for example, what happens to sales if I increase ad spending by 10%?