Need of Machine Learning to Predict Happiness: A Systematic Review

Abstract

Happiness is a current important subject of study in psychology and social science because it affects people's day-to-day lives, thoughts and feelings, work habits, and interactions with society and family. There are a number of challenges in Computer Science and Machine Learning to predict happiness index using prediction techniques. This study presents a systematic review using PRISMA style for happiness prediction. During the Literature survey, it was found that many predictive models whether statistical or Machine Learning was designed to predict happiness index but a major emphasis on research remains focused on the factors that are listed in World Happiness Report, i.e., real Gross Domestic Product per capita, social support, healthy life expectancy, freedom to make life choices, generosity and perceptions of corruption. The factor influencing happiness varies due to personal differences, age group and location variation. According to Gallup Poll, the general annual sample for each country is 1,000 people i.e., approximately 0.007% population participated in happiness index measurement. The purpose of this study is to discover and describe new factors related to psychology like stress and emotions, location-based and age group. It is observed that there is a requirement to develop a Machine Learning predictive model which works on psychological factors like mental health, depression, stress, physical well-being, safety, leisure time available, and suicidal ideation in addition to economic factors used in World Happiness Index and by targeting a large sample size of populations.