Analyzing the Sensitivity of Neighborhood size on Prediction Quality using  Collaborative Filtering

Babu Reddy.M

ABSTRACT Traditional statistical techniques are to be extended to handle the high dimensional data during machine learning and prediction tasks. In many of the applications, the size of data will be exorbitant: from retail marketing data to biomedical images and natural resource information. For such high-dimensional statistical explorations, Feature selection is essential. As the dimensionality of data gets increased day by day, there is an essential need to follow new mathematical approaches to deal with the high dimensional data. Unlike traditional statistical approaches, in machine learning and prediction scenarios, the risk minimization is considered as a key issue. In view of this, all the mathematical methods in connection with machine learning are moving around the understanding of the performance of learning theory. In these lines, this paper highlights the dynamics associated with the collaborative filters which are very much useful in prediction tasks. An effort has been made to analyze the relationship among the percentage of training samples used, learning rate applied and neighborhood size, especially in supervised learning environment. Bench mark micro array data set of English Alphabets has been used in the experimental setup.

Title:	Analyzing the Sensitivity of Neighborhood size on Prediction Quality using Collaborative Filtering
Author Name:	Babu Reddy.M
Abstract:	ABSTRACT Traditional statistical techniques are to be extended to handle the high dimensional data during machine learning and prediction tasks. In many of the applications, the size of data will be exorbitant: from retail marketing data to biomedical images and natural resource information. For such high-dimensional statistical explorations, Feature selection is essential. As the dimensionality of data gets increased day by day, there is an essential need to follow new mathematical approaches to deal with the high dimensional data. Unlike traditional statistical approaches, in machine learning and prediction scenarios, the risk minimization is considered as a key issue. In view of this, all the mathematical methods in connection with machine learning are moving around the understanding of the performance of learning theory. In these lines, this paper highlights the dynamics associated with the collaborative filters which are very much useful in prediction tasks. An effort has been made to analyze the relationship among the percentage of training samples used, learning rate applied and neighborhood size, especially in supervised learning environment. Bench mark micro array data set of English Alphabets has been used in the experimental setup.
Cite this article:	Babu Reddy.M , " Analyzing the Sensitivity of Neighborhood size on Prediction Quality using Collaborative Filtering " , International Journal of Application or Innovation in Engineering & Management (IJAIEM), Volume 3, Issue 7, July 2014 , pp. 237-241 , ISSN 2319 - 4847.
Full Text [PDF] Home