Viscovery SOMine

Category Intelligent Software>Data Mining Systems/Tools

Abstract Viscovery SOMine is an advanced data mining system based on Self-Organizing Maps (SOM). SOMine can be used for exploratory data mining, statistical analysis, profiling and segmentation, visual cluster analysis, and classifications.

SOMine 5.2 features and functions --

Viscovery SOMine combines classical statistical methods and SOMs in a system for explorative data mining and predictive modeling.

The robust, high-performance SOM technology is used for representing and visualizing data distributions that may contain thousands of variables (e.g., in text mining applications) and millions of data sets.

In contrast to traditional data mining systems, Viscovery provides an intuitive approach to data, so that even users who do Not have advanced statistical knowledge can easily understand and use the analytical models.

Viscovery thus embodies a unique visual approach to data which facilitates understanding and communication of analytical models.

All tasks, including importing the data, pre-processing and treatment of outliers, and defining segments and measures, are guided by workflows. Completed models can be modified and applied again to new projects.

Viscovery SOMine provides easy-to-use tools for data exploration, identification of dependences, visual cluster analysis, segmentation and classification, as well as a number of classical statistical functions, such as descriptive statistics, group profiles, correlation analysis, PCA, histograms, scatter plots, and others.

The visual interface allows direct, context-sensitive access to the original data records underlying the maps and their complete statistics at any point in the analytical workflow (Expert and Enterprise editions - see below...).

SOMine is an affordable solution for explorative data mining, cluster analysis and classification.

It is an alternative to the enterprise solution “Viscovery Profiler” which has the additional capabilities of providing database interfaces and the capabilities to automate model creation and application and to integrate models in real-time environments.

Products Main Functions/capabilities include:

Data Preprocessing --

1) Definition of nominal variables;

2) Removals (including conditional removal using a combination of several attributes);

3) Statistical and deterministic sampling and over sampling;

4) Replacements;

5) Transformations;

6) Sampling to accelerate model generation;

7) Computation of new variables using a custom formula language;

8) Renaming of attributes, attribute descriptions; and

9) Treatment of outliers.

Technology / Self-Organizing Maps --

1) Generation of multivariate data order;

2) Reduction to a 2-dimensional data representation and its visualization (attributes, clusters, segments, U-matrix, frequency, quantization error, group profile);

3) Possibility to define the influence of individual attributes on the data ordering using priorities;

4) Automatic compensation of correlations in the data; and

5) Well-defined treatment of missing values.

Definition of Segments --

1) Automatic cluster methods (SOM-Ward, Ward, SOM-Single-Linkage);

2) Instant retrieval through selections on the model;

3) Display of original data records that correspond to selected areas in the model;

4) Segment definition by selection using a mouse;

5) Precise segment descriptions using Group Profiles (statistical description of segments) -- the possibility of documentation is included;

6) Assignment of concrete actions to segments;

7) Management of actions (per project, can be imported and exported);

8) Arbitrary number of segmentations in each single model; and

9) Business rules for segments can be defined using formulas.

Assignment of Data Records to Segments and Actions --

1) The assignment of each individual customer to a segment and action as well as the evaluation of business rules (formulas).

2) The result is a flat table containing the segment, the action, and results of the formulas for each data record.

Evaluation of Campaigns --

1) Charts for visual and quantitative evaluation of models and campaigns.

2) All model attributes and all values from each campaign are available for visual inspection.

Products Additional Features/capabilities include:

Workflow-Orientation --

1) Optimized workflows lead you through the application (Create Data Mart, Create Model, Apply Model, and Evaluate).

2) Workflows can be processed automatically on an optional basis.

3) Integrated project documentation by description of completed workflows.

4) A dedicated workflow supporting the decision process for segment definition (including documentation).

Visual Representations --

1) Viscovery SOM visualization including segments and U-matrix;

2) Histograms; and 3) Charts of other important parameters (e.g. segment comparison, group profile, etc.).

Statistical Information (available with the Expert and Enterprise Editions of this product) --

1) Available for each workflow step - context sensitive;

2) Data record browser (subsets can be selected in the histogram);

3) Descriptive statistics;

4) Correlation matrix;

5) Histograms and outlier treatment;

6) Principal Components Analysis (PCA) (attributes can be selected);

7) Frequency table;

8) Box plots; and

9) Scatter plots.

Reports --

1) Instant reports for each workflow step as well as the entire workflow is available at anytime;

2) Reproducibility of completed workflow steps by automatic reporting; and

3) Reports include all documentation and descriptions entered by the end-user.

Usability --

1) Simple operation because the user is shielded from the technology core;

2) Can be operated by business users;

3) Info pop-ups in the workflow and in the map (SOM visualization), attribute descriptions are also included;

4) The possibility to define labels and paths over the map;

5) Branching of workflows and copying of workflow steps;

6) Project management in project directories and “clean directories” function;

7) Helpful context menus at numerous places;

8) Tables can be sorted; and

9) Combinations of visible windows, attribute selections, etc. including their sizes and appearances can be stored in “arrangements”.

Basic Edition of SOMine --

1) Up to 100,000 data records with up to 100 variables can be processed.

Expert Edition of SOMine --

1) All features and functionalities of Viscovery SOMine.

2) Direct access to the original data records including their statistical information through-out the model and context sensitive info for each workflow step.

Enterprise Edition of SOMine --

1) All features and functionalities of Viscovery SOMine Expert Edition.

2) Unlimited number of data records and an unlimited number of variables can be processed.

System Requirements

Contact manufacturer.


Manufacturer Web Site Viscovery SOMine

Price Contact manufacturer.

G6G Abstract Number 20115R

G6G Manufacturer Number 102900