Design and evaluation of logistic regression model for pattern recognition systems

In this paper, an attempt is made to design pattern recognition systems using logistic regression model and few mapping functions are proposed for the same. The performance of proposed logistic regression model and mapping functions are assessed by evaluating the model using design of digital circuits, standard datasets from UMASS database and datasets pertaining to wireless sensor network applications. It is observed that for a majority of cases, the recognition accuracy is enhanced on using proposed mapping functions for both binary and multi class pattern classification problems.