Basic Concept of Classification

Question 1

given a collection of records, each record by characterized by a tuple (x,y), where x is the attribute set and y is the label set.

Answer 1

Classification

Question 2

6 CLASSIFICATION TECHNIQUES

Answer 2

1. Decision Tree Based Methods
2. Rule-based Methods
3. Memory Based Reasoning
4. Neural Network / Deep Learning
5. Naive Bayes and Bayesian Belief Network
6. Support Vector Machines

Question 3

DECISION TREE: INDUCTION:

Answer 3

Training set is being inducted to train a model (Learning Algorithm -> Learn Model),

the model will be able to form a decision tree,

and we can apply the model to deduct from the test set.

Question 4

is a type of algorithm that uses attributes to split the data recursively, till each split contains only a single class.

Answer 4

Hunt’s Algorithm

Question 5

4 TYPES OF ATTRIBUTES

Answer 5

1. Binary
2. Nominal
3. Ordinal
4. Continuous

Question 6

2 TEST CONDITION FOR NOMINAL ATTRIBUTE

Answer 6

• Multi-way Split – use as many partitions as distinct values.
• Binary Split – divides values into two subsets.

Question 7

2 TEST CONDITION FOR ORDINAL ATTRIBUTE

Answer 7

• Multi-way Split – use as many partitions as distinct values.
• Binary Split – divides values into two subsets and preserve order property among attribute values.

Question 8

is an approach to getting the best split where nodes with homogenous class distribution are preferred.

Answer 8

Greedy Approach

Question 9

Formula for General Framework when finding the best split.

Answer 9

M0 is the value of the parent.

M12 is Node 1 * Node 2

M34 is Node 3 * Node 4

Gain = M0 - M12 VS M0 - M34

Question 10

3 WAYS TO MESURE NODE IMPURITY

Answer 10

1. Gini Index
2. Entropy
3. Classification Error

Question 11

is a measure of how often a randomly chosen element from the set would be incorrectly labeled if it was randomly labeled according to the distribution of labels in the subset.

Answer 11

Gini Impurity / Index

Question 12

measures homogeneity of a node uncertainty of a random variable or information content of a message.

Answer 12

Entropy

Question 13

measures misclassification made by a node.

Answer 13

Classification Erro

Question 14

3 STOPPING CRITERIA FOR TREE INDUCTION

Answer 14

1. Stop expanding the node when all the record belongs to the same class.
2. Stop expanding a node when all the records in the node have the same attribute.
3. Early Termination Criteria.

Question 15

4 ADVANTAGES OF DEVISION TREE BASED CLASSIFICATION

Answer 15

1. Inexpensive to Construct
2. Extremely Fast at Classifying Unknown Records
3. Easy to Interpret for Small-Sized Trees
4. Accuracy is Comparable to other classification techniques.