Classification

Question 1

Classification

Answer 1

The process of categorising data into predefined classes based on their attributes, characteristics, or features.

Question 2

2 Approach of Building Model

Answer 2

Induction - Learn model
Deduction - Apply model

Question 3

4 Types of Classification

Answer 3

Binary Classifier
Multiclass Classifier
Multilabel Classifier
Multioutput Classifier (Multioutput–multiclass classification)

Question 4

Class vs Label

Answer 4

Class: A group or category of data points with similar characteristics or attributes.
Label: A name or identifier assigned to a particular data point that indicates the class to which it belongs.

Question 5

Batch Learning

Answer 5

Training the model on the entire dataset at once; No update

Question 6

2 Design Issues of Decision Tree Induction

Answer 6

Splitting criterion: Attribute test condition to divide records into smaller subsets
Stopping criterion: All records belong to the same class, Early termination

Question 7

1 Approach Used in Splitting Criterion

Answer 7

Greedy approach - Nodes with purer class distribution are preferred

Question 8

3 Step to Find the Best Split

Answer 8

Compute impurity measure (P) before splitting
Compute impurity measure (M) after splitting
Choose the attribute test condition that produces the highest gain.

Question 9

4 Advantage of Decision Tree Based Classification

Answer 9

Relatively inexpensive to construct
Easy to interpret for small-sized trees
Fast at classifying unknown records
Versatile

Question 10

2 Disadvantage of Decision Tree Based Classification

Answer 10

Each decision boundary involves only a single attribute
Instability

Question 11

3 Measures of Node Impurity

Answer 11

Gini Index - min. = 0; max. = 0.5
Entropy - min. = 0; max. = 1
Misclassification error - min. = 0; max. = 0.5