INVESTOR’S PERCEPTION OF MUTUAL FUND INVESTMENTS WITH SPECIAL REFERENCE TO HDFC MUTUAL FUND-CHENNAI

Mutual fund model estimation has been a common research steam for solving various
applications like portfolio selection, prediction, performance evaluation and style analysis
(classification) of mutual funds. All these applications require one to identify the set of most
relevant features and a basic function estimation model to be built using these features, which
can then be used to classify, estimate or predict values.
Feature selection for mutual funds has been done in the past on the basis of expert
domain knowledge using computational feature extraction techniques like principal component
analysis or by applying subset selection methods to statistical techniques. Similarly function
approximation has traditionally relied on expert domain knowledge statistical method (like linear
and non linear regression) and linear programming based models. In recent years, approaches
based on machine learning such as neural networks genetic algorithms, support vector machine,
have become popular tools. Unlike most computing algorithms these approaches do not require
the mathematical model to be specified beforehand. Genetic algorithms are used for finding
solutions to optimisation problems