Link to the previous post : https://statinfer.com/204-4-9-model-bias-variance-tradeoff/
We always build and train a model with a training dataset. With default parameters the fitting process optimizes the training data as well as possible. Introducing, a whole different sample to pre-built model won’t replicate the accuracy we expect form the model.
One way to solve this problem is cross validation. We can divide our dataset into training and testing samples. After fitting the model with training sample we can validate the accuracy on test sample. This method is called cross validation.
Commonly used Cross Validation techniques:
- Hold-Out data Cross Validation
- K-fold Cross Validation
- 10-fold Cross Validation
- Bootstrap Cross Validation
Holdout Data Cross Validation
- The best solution is out of time validation. Or the testing error should be given high priority over the training error.
- A model that is performing good on training data and equally good on testing is preferred.
- We may not have the test data always. How do we estimate test error?
- We take the part of the data as training and keep aside some potion for validation. May be 80%-20% or 90%-10%
- Data splitting is a very basic intuitive method
LAB: Holdout Data Cross Validation
- Data: Fiberbits/Fiberbits.csv
- Take a random sample with 80% data as training sample
- Use rest 20% as holdout sample.
- Build a model on 80% of the data. Try to validate it on holdout sample.
- Try to increase or reduce the complexity and choose the best model that performs well on training data as well as holdout data.
#Splitting data into 80:20::train:test X_train, X_test, y_train, y_test = train_test_split(X,y, train_size = 0.8)
#Defining tree parameters and training the tree tree_CV = tree.DecisionTreeClassifier(criterion='gini', splitter='best', max_depth=20, min_samples_split=2, min_samples_leaf=1) tree_CV.fit(X_train,y_train)
DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=20, max_features=None, max_leaf_nodes=None, min_samples_leaf=1, min_samples_split=2, min_weight_fraction_leaf=0.0, presort=False, random_state=None, splitter='best')
#Training score tree_CV.score(X_train,y_train)
#Validation Accuracy on test data tree_CV.score(X_test,y_test)
Improving the above model:
tree_CV1 = tree.DecisionTreeClassifier(criterion='gini', splitter='best', max_depth=10, min_samples_split=30, min_samples_leaf=30, max_leaf_nodes=30) tree_CV1.fit(X_train,y_train)
DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=10, max_features=None, max_leaf_nodes=30, min_samples_leaf=30, min_samples_split=30, min_weight_fraction_leaf=0.0, presort=False, random_state=None, splitter='best')
#Training score of this pruned tree model tree_CV1.score(X_train,y_train)
#Validation score of pruned tree model tree_CV1.score(X_test,y_test)
The model above is giving same accuracy on training and holdout data.
The next post is about k fold cross validation.
Link to the next post : https://statinfer.com/204-4-11-k-fold-cross-validation/