Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
0% found this document useful (0 votes)
13 views

SLIQ_Algorithm_with_Example

SLIQ (Supervised Learning In Quest) is a decision tree induction algorithm designed for efficient classification of large datasets. It utilizes a pre-sorting technique and a class list to quickly build classification trees by selecting the best splits based on the Gini index. An example demonstrates how SLIQ processes a dataset to create a decision tree with optimal splits.

Uploaded by

BINDUANN THOMAS
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
13 views

SLIQ_Algorithm_with_Example

SLIQ (Supervised Learning In Quest) is a decision tree induction algorithm designed for efficient classification of large datasets. It utilizes a pre-sorting technique and a class list to quickly build classification trees by selecting the best splits based on the Gini index. An example demonstrates how SLIQ processes a dataset to create a decision tree with optimal splits.

Uploaded by

BINDUANN THOMAS
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

SLIQ Algorithm with Example

What is SLIQ Algorithm?


SLIQ stands for Supervised Learning In Quest. It is a decision tree induction algorithm
developed to handle large datasets efficiently, especially for classification tasks. SLIQ
improves earlier methods by using a pre-sorting technique, building classification trees
quickly without re-sorting at every node, and maintaining a class list to track data.

Steps in SLIQ:
 Pre-sort each attribute once.
 Initialize a 'class list' with record ID, class label, and node number.
 Choose the best split based on Gini index using sorted attributes.
 Split the class list based on the selected attribute.
 Repeat recursively for each child node.
 Stop when nodes are pure or meet stopping conditions.

Example Dataset:
ID Age Income (K) Class

1 22 25 No

2 45 50 Yes

3 27 30 No

4 35 45 Yes

5 50 60 Yes

Attributes Pre-Sorted:

Age: 22, 27, 35, 45, 50

Income: 25, 30, 45, 50, 60

Class List Created:

ID Class Node

1 No 0

2 Yes 0
3 No 0

4 Yes 0

5 Yes 0

Finding the Best Split:


Split based on Age at 30 gives the best Gini index (perfect split).

Gini Index Calculation:


Split at Age <= 30:

Left Group: IDs 1, 3 (Class No)


Right Group: IDs 2, 4, 5 (Class Yes)

Gini(Left) = 0
Gini(Right) = 0
Weighted Gini = 0

Final Decision Tree:


[Root: Age <= 30?]
├── Yes (<=30) → Class: No
└── No (>30) → Class: Yes

Summary Table:
Split Condition Left Group Right Group Gini Split Remarks

Age <= 22.5 1 record (No) 4 records (3 0.3 Not best


Yes, 1 No)

Age <= 30.0 2 records (No) 3 records (Yes) 0.0 Best Split

Age <= 40.0 3 records 2 records (Yes) >0 Not better


(mixed)

You might also like