Genetic Algorithm Based Optimized Ensemble Model for the Prediction of Happiness Index -
Happiness is one of the most important goals for humans. “Aristotle considered his euphoria, translated as ‘happiness’, to be ultimate and greater goal of humans “. Happiness is considered to be associated with psychological functions. Happiness is an extremely treasured part of life, greater to all values like cash, health or anything else. Therefore, happiness is the first and foremost desirable condition among all people on the face of the earth, and other aims may only be seen as components for determining happiness. By an evolutionary view, happiness is often seen as a psychological reward for everything. Through the years, considerable advances are made within the basic understanding and treatment of psychological state. However, treatment isn't just fixing what's not right within us; it also involves feeding what's greatest within ourselves. Happiness could be a positive concept that's very important in maintaining a healthy lifestyle , yet there are only few studies that show the usefulness of happiness and how to measure them.
Any single model cannot capture the whole essence of the problem and also fails to make accurate predictions. So it's often seen that using multiple models have solved this issue significantly by improving the accuracy provided by any single model. So we develop a single model by aggregating all the base models by using various techniques like bootstrapping, averaging, weighted averaging. In this project we use three base models Support Vector Machine(SVM), K-Nearest Neighbour and decision trees and optimize them using the genetic algorithm
GAs are a heuristic arrangement search or optimization techniques, initially roused by the Darwinian standard of evolution theory. We portray how to develop a GA and the principle strands of GA hypothesis before theoretically distinguishing potential utilizations of GAs to the investigation of optimization. A GA utilizes a profound work on the theoretical cycle of generation to develop the best optimal search solutions in the given space. Every GA has been run on real coded genetic (binary valued) chromosomes string of particular gene length (L). Every binary chromosome acts as a solution to the problem and generates the fitness value, which acts as measurement for a particular solution to represent its betterness for specific problems.