Skip to content

Classification Notes

Overview

  • Predictive modeling approximates a mapping function which predicts the class or category far a given observation.

Linear Model

                +-----------+        
X(x1,x2) ====> |    Model  | ====> Classified(x1, x2)  
                +-----------+        features
- where : - x1 : classe1/label1/target1 - x2 : classe2/label2/target2

Ex: Diabet classification model

- high bloods sugar levels  <=> old age
        |                           |
        v                           v
- insulin resistance        <=> less exercise 
        |                           |
        v                           v
- diabet task               <=> diabetic risk
- labels to classify : [no disease, disease]

Prediction(scoring) - if new person (classified input) is below the line (model) then we estimated it its no disease class according to previous data locationn

Type of classification models

  • Binary classification
  • Multi-Class classification
  • Multi-Label classification
  • Imbalanced classification

References

Notes From : The Complete Self-Driving Car Course Applied Deep-Learning - Udemy - Notes