Prediction and Classification of Human Development Index Using Machine Learning Techniques

Keeping pace with the global advancement and tracking the development of other countries is very important for Bangladesh, also monitoring the less developed regions is needed to take necessary development steps. To facilitate that, we have created a system which classifies Human Development Index of 186 countries. Using various machine learning algorithms we were able to forecast the Human Development Index of those regions till 2025 which will be updated every year. As south Asian countries are quite similar to Bangladesh in terms of health, economy, education, we have emphasized more on these countries. In the first part, we have tested different time series analysis models, AutoRegressive(AR) model,Autoregressive Integrated Moving Average(ARIMA) model, Weighted Moving Average(WMA) model, and Prophet model for forecasting. The Mean Absolute Percentage Errors(MAPE) were 0.04% for AR, 0.18% for WMA, 0.04% for ARIMA, and 0.32% for Prophet model. In the second part, we have tested K-means and hierarchical clustering algorithms to group values of HDI indicators (gross national income, life expectancy, expected years of schooling, mean years of schooling ), where the four labels are ’low’ which indicates countries with lower HDI, ’medium’ denotes countries which are still developing, ’high’ specifies countries which are moderately developed and ’very high’ represents the countries with the highest HDI. In the variance calculation, values found for K-Means was 0.0125 and hierarchical clustering was 0.00752 respectively.