Global Land Temperature Prediction by Machine Learning Combo Approach

The Global Land Surface Temperature is the radiative skin temperature of ground, depending on factors, which includes the albedo, the vegetation covers and the soil moisture. To predict the changes in temperature in a particular region is becoming increasingly important to capture the future trends in that region. Machine Learning is a specialized branch of Artificial Intelligence (AI), which gives computers the power to learn and make predictions from the data, without being explicitly programmed. In this work, Ensemble Approach for Global Land Temperatures (EAGLT) is proposed. This approach will help to predict the temperature, which is of great requirement as the problem of global warming is increasing day by day. Temperatures are collected from different cities and prediction is done using this approach. The proposed ensemble approach is based on three models which provide good performance in terms of model evaluation parameters like Correlation, Accuracy, R-Squared (R-2), Root mean square (RMSE) and Total Time to detect the predicted temperatures. Cross Validation is performed on the best performing models to check the robustness of these selected models.