4. Instance Attribute 1 Attribute 2 Attribute 3 Class 1 point T Y 2 T 6 Y 3 T 5 N 4 F 4 7 N 3 N B 7 5 Figure 1: Training Dataset Review the table labeled Figure 1 Training Dataset. Assume that we want to use a decision tree for modeling the data. Using the misclassification error as the measure of node impurity, which attribute, between Attribute 1 and Attribute 2. provides the best split? What is the misclassification error rate for the split? Attribute 2 provides the best split 4/9 is the error rate for the split Attribute 1 provides the best split 4/9 is the error rate for the split Attribute 1 provides the best split 2/9 is the error rate for the split Attribute 2 provides the best split 2/9 is the error rate for the split 5. Instance Attribute 1 Attribute 2 1 point Attribute 3 Class 1 T T 1 Y 2 T T 6 Y 3 T F 5 F F 4 5 T 7 6 T 3 ZZZZZ 7 F 8 8 7 9 F 5 Figure 1: Training Dataset Review the table labeled Figure 1: Training Dataset. Assume that we want to use a decision tree for modeling the data, Using the Gini Index as the measure of node impurity, which attribute between Attribute 1 and Attribute 2, provides the best split? What is the Gini Index for the split? Attribute 1 provides the best split. Gini Index is 0.4889 Attribute 1 provides the best split Gini Index is 0.3444 Attribute 2 provides the best split. Gini Index is 0.3444 Attribute 2 provides the best split. Gini Index is 0.4889 1 point 6. Your friends have trained a model to detect human faces in an image, but the model has high training and high generalization errors. To address the high training and high generalization errors in the model which suggestion would you provide to your friends? The model is overtit, which means it has low blas but high variance so variance should be reduced. The model is underfit which means it has high bias but low variance, so bias should be reduced. The model is underfit, which means it has low bias but high variance, so variance should be reduced. The model is overfit, which means it has high bias but low variance, so bias should be reduced