Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 11

Dynamic Pruning of Random Forest

Under the esteemed supervision of Mrs.


Nagarjuna Devi
Team
Y. Poorna Durga
members:
P. Saimadhu

Dynamic Pruning of Random Forest Classifier

Introduction
Existed Methods
Proposed Method
Requirements analysis
Implementation
Results

Introduction

Random forest overview


Random forest Algorithm
Applications of Random Forest Classifier

Existed Methods

Static Pruning Methods


Methods based on diversity measures.
Methods based on search algorithms.
Methods based on clustering of classifiers.
Methods based on heuristic rules.

Proposed Method

Dynamic pruning

Requirements analysis

Programming Languages : Python


Packages Required : NumPy
Tools : Weka
IDE : Stani's Python Editor(SPE)
Data sets : Titanic, Qualitative Bankruptcy , Cancer
detection , USA voting

Implementation
Information Gain module
Decision Tree module
Random Forest module( Including proposed method)

Results
Cancer Detection with 10 attributes including target
attribute
Weka accuracy average of 10 results : 70.04 %
Proposed method average of 10 results : 74.20 %

Results
Qualitative Bankruptcy with 7 attributes including target
attribute
Weka accuracy average of 10 results : 92.04 %
Proposed method average of 10 results : 95.32 %

Results
Time Complexity : For building a tree is O(mnlogn)
Time Complexity : For building M trees O(M(mnlogn)
Where :
M : Number of trees.
m : Number of attributes.
n : The number of instances in the training data.

Thank you

You might also like