Scenario:

     Assume your mailing/email list has 2 million names, and hidden in this list, conservatively assume that there are only 8,000 (0.4%) potential customers for your product. If your current predictive models find these potential customers with a 60% accuracy, it means you can correctly identify only 4,800 of the 8,000 potential customers. With EZPredict, you can get similar or higher accuracy than Logit/Logistic Regression models, Neural Networks, or Tree based models. Even a 3% higher classification accuracy implies a new opportunity to market to 240 more potential customers. Assuming US $250 in annual profit from each of these customers, you could stand to gain US $60,000.00 annually by using EZPredict for your predictive modeling!

• WHY USE EZPredict?

  GET MORE ACCURATE PREDICTIONS THAN LOGIT, DISCRIMINANT ANALYSIS, NEURAL NETWORKS, AND TREE MODELS. AND MORE...

  • Easy-to-use methodology from leading inventors of Marketing Analytics Methodology.
    
    
  • Choice of up to six models: Get predictions from 6 to 10 different models. Use whichever meets your needs!!
    
    
  • No expertise in modeling necessary. Don’t need SAS, SPSS, or PhD in Analytics to use this site. No Hassles of weighting variables, understanding deciles, quartiles, GINI Index, or Likelihood Ratios.
    
    
  • Multiple Data formats: Read from your EXCEL files, Databases, or delimited Text Files.
    
    
  • Large-data friendly: Upload Text files with 99% size reductions using WinZip 256-bit password-protected encryption. (A 1GB text file can be reduced to 10MB using WinZip).
    
    
  • Data Security using SSL technology: Your data gets transferred using secure 128-bit encrypted SSL technology.
    
    
  • Convenient pricing: Only score databases if the model results meet your requirements. Nominal fees for model development and database scoring.
    
    

• EXAMPLES OF SUPERIOR PREDICTIONS OF EZPREDICT COMPARED TO OTHER TECHNIQUES
  
  
  
  • Services Industry - Brand Perceptions
    (3-category, High-Medium-Low) Dependent Variable – Holdout sample
    
    

    Prediction Method	Overall Correct Classification	Average of Correct Classifications	Geometric Mean of Correct Classifications
    Tree Model	68%	48%	0%
    EZPredict	68%	54%	44%

    
    
    
  • Telecom Industry – Direct Marketing Example 1
    (4-category Market-Segment Dependent Variable) – Holdout sample
    
    

    Prediction Method	Overall Correct Classification	Average of Correct Classifications	Geometric Mean of Correct Classifications
    Tree Model	67%	69%	65%
    EZPredict	70%	71%	70%

    
    
    
  • Greetings Industry – Individual level Conjoint Analysis
    Holdout Profiles
    
    

    Prediction Method	Overall Correct Classification	Average of Correct Classifications	Geometric Mean of Correct Classifications
    Tree Model	Can't estimate	Can't estimate	Can't estimate
    Logit Model	59%	55%	55%
    EZPredict	74%	53%	66%

    
    
    
  • Telecom Industry – Direct Marketing Example 2
    (4-category Market-Segment Dependent Variable) – Holdout sample
    
    

    Prediction Method	Overall Correct Classification	Average of Correct Classifications	Geometric Mean of Correct Classifications
    LOGIT Model	62%	58.5%	48.2%
    EZPredict	69.4%	69.3%	67.5%