Abstract:
In current situation, each user is generating gigabytes of data each day. The quantity of data is so huge that we have dependent on machine learning models to capture import data / to capture patterns from data that can be useful for future prediction.
To make use of machine learning models, we required continuous computational power. The system will breakdown if connection to computation breaks. When we try to reduce the computational power then we have to compromise with accuracy.
This thesis represents the idea of “Breast Cancer Classification using Logical Analysis of Data”
The thesis presents a review of the basic concepts of the Logical Analysis of Data& put the focus on the various methods those can be used in different components of LAD. Binarization methods includes different methodologies to covert complex attributes into binary.
The main feature of the Logical Analysis of Data (LAD) is to find minimum set of features those can cover all the observation with approaches like coefficient correlation, threshold count, set covering. The decision tree classifier has been used to find the patterns in the observations.
This thesis also looks for a hardware implementation of classifier so that continuous connectivity is no longer needed.
