Please enable JavaScript.

Coggle requires JavaScript to display documents.

Ch17-18 (ROC Curve: receiver operating characteristics, curving up and to…

- - - - TP+TN/All Cases= accuracy of model (proportion of decisions made that are correct) considers all quadrants but assumes negatives and positives are of equal importance
    - - F1 Score= harmonic mean of Positive Predictive Value and True Positive Rate 2TP/ 2TP+FP+FN
  - - - Negative Predictive Value= TN/TN+FN
        
        True Negative Rate= TN/TN+FP
        
        False Positive Rate= FP/FP+TN
- - - - understanding a model's performance and business context is key
        
        FVE Binomial (fraction variance explained) is equivalent to R2 value
        
        OR how far off percent wise the model is from fully explaining who will be readmitted
        
        Decision tree algorithm is easiest conceptually and most effective
        
        Decision Tree: 1. find most predictive feature and place at root of tree
        
        Split feature into 2 groups at point where feature is most homogenous
        
        1 more item...
        
        FVE Binomial expresses how much variance in the dataset has been explained