• Implement proprietary data mining technologies
• Ideal for non-linear, non-Gaussian, noisy, chaotic data
• Superior data separation by hidden projection
• Identify topological pattern of sample data
• Include modules for PCA & Fisher method
• Parameter estimation by neural networks
• Network optimization by genetic algorithms
• Interactive modeling & simulation tools 
• Offer tools for feature selection and reduction
• Predict optimal condition and operation zone
• Complete data conditioning/analysis tool set
• Calculate target-factor sensitivity curves
• Visualize target/factor association and correlation
• Read data from Microsoft Excel and text files
• Runs on PC/Windows and NT

MasterMiner™ is a complete software suite for intelligent data mining.  It implements Zaptron's proprietary technologies in data mining, pattern separability, projective geometry, pattern topology, feature multiplicity and selection, and sub-space modeling . It is an ideal tool kit for general process optimization and design of experiment applications where one expects maximum return by discovering meaningful new correlations, patterns and trends. MasterMiner has been used in manufacturing, environmental, chemical, material, petroleum, semicoductor, pharmaceutical, metallurgical, automobile, steel and financial applications.

Data Loading and Conditioning
Read data in Excel and text format
Auto-check of data format and data collinearity
Offer local views of data
Capability to delete outliers and noise
Virtual sampling by random data appending

Data Separability Test
Determine data separability by geometric method

Statistical Analysis
Linearity test
Compute table of nearest neighborhood data points
Operations to find best matching points
Identify topologic type of data patterns (inclusive, 2-sided)
Factor-factor association analysis
Target-factor correlation analysis

Feature Selection
Selection by pattern classification
Offer diferent methods for different data patterns
Interactive feature reduction

Model Building
Modeling in feature sub-space
Inequalities as boundary to segment hyperspace
Linear equations in two principal directions
Offer control advisory
Display 4-dimensinal cross-sectional maps

Neural Networks Included
Define nodes and network configuration
Display training error-time plot
Predict target values from subspace data
Estimate factor values
Optimize analystical solutions by a genetic algorithm
Improved performance with boundary conditions
Factor sensitivity analysis
Dispaly color maps for optimal operating zone
Dispaly all nods and weighted computed from data
Allows extrapolation for maximum economic benefit

AutoBox Tool
Effective tool to mine data in hyperspace
Graphically segment hyperspace int subspaces
Cutting tool to manually eliminate data from a subspace

Advantages
No extra equipment added
No tedious experiments needed
No undesired interruption to production

Order Information
Zaptron Systems, Inc.
415 Clyde Avenue, Suite 108
Moutain View, CA 94043 USA
Tel: 650-966-8700, Fax: 650-966-8780
Email: sales@zaptron.com
Web: www.zaptron.com