DOI:10.20894/IJCNES.
Periodicity: Bi Annual.
Impact Factor:
SJIF:5.217
Submission:Any Time
Publisher:IIR Groups
Language:English
Review Process:
Double Blinded
Data Analysis and classification is widely used in many applications. Classification is one of the data mining methodologies which help to assign data in a collection to target classes. The objective of this work is to develop a rough set based Neuro-fuzzy inference system for decision making with high predictive accuracy. In this work, we have proposed an enhanced technique to incorporate the concept of rough set theory with Neuro-fuzzy system to minimize the uncertainties of the data set by removing the redundant attributes. The goal of classification is to forecast accurately the target class for each data case. Huge volumes of data will usually have a large number of attributes and there will be lot of missing and invalid attributes. To overcome these shortcomings due to invalid or missing attributes we incorporate rough set approximation concept with Neuro-fuzzy. Thus, the proposed system will increase the accuracy of the classification even with missing attributes and the rules generated thereby resulting in efficient classification of decision attributes based on the condition attributes. The proposed methodology can be used for forecasting and prediction techniques such as clinical diagnosis, stock market prediction, weather forecasting, landslide prediction and so on.
