DDE-1/Classification.md

@@ -10,6 +10,7 @@ In the previous week, we discuss one of the predictive learning tasks, [regressi
 
 <div align="center">
 <img src="uploads/c8788ae7ed388aac4da978e352a6e1e6/lr1.png"  width="600">
+
 Outline of the logistic regression model
 </div>
 
@@ -101,6 +102,18 @@ The recursive algorithm is stopped when either of the followings is true:
 
 [CART]( https://www.analyticssteps.com/blogs/classification-and-regression-tree-cart-algorithm) is the tree generator used in [scikit learn]( https://scikit-learn.org/stable/modules/tree.html#tree-algorithms). It can create both Classification And Regression Trees, --see the initials. Constructed trees are binary: each parent has two child nodes. The split is done based on the twoing criteria. Herein, we search for the best homogeneity for the child nodes via [Gini index]( https://en.wikipedia.org/wiki/Gini_coefficient). Once the tree is created, it is pruned via cost-complexity balancing. In the regression, the split criterion is based on the minimization of the squared errors. 
 
+## Model evaluation
+
+Once the models of interest are trained and ready, we test them by using various tools:
+
+
+<div align="center">
+<img src="uploads/89b65167fd3b61ce5bd3779b7db794cb/lr2.png"  width="600">
+
+Outline of model testing with LogLoss, Confusion Matrix, ROC and PR curve
+</div>
+
+You may find the details in the lecture notes. 
 
 ## Additional references