Nigeria Poverty Prediction Using Statiscal Analysis and Machine Learning

Abstract: A multitude of deficiencies are referred to be poverty, including the inability to supply basic needs like clothing, food, clean water, and shelter. In the modern world, this also includes having access to education and health care. Numerous researchers make a great effort to comprehend, analyse, and forecast poverty using various methods. As a result, the Nigerian government was provided access to these other methods, which are ineffective. The traditional method of prediction in Nigeria involves site surveys, which are costly, labour-intensive, and a waste of time and energy before actual results are obtained. The predictions are also not accurate. The primary issue and barrier to making well-informed policy decisions and efficiently distributing resources in those areas that need the greatest assistance in Nigeria is the lack of trustworthy data on poverty. To determine which prediction model is more accurate for predicting poverty in Nigeria, we will attempt to compare many of them. The use of large data for measurement has been made possible by strategies based on machine learning development. Additionally, in this thesis, we will compare the prediction accuracy using models based on decision trees, binary logistics regression, and random forests.