Applying Predictive Analytics to Climate Change: Predicting Temperature Rise Using Human Behavior Alternative Data

Climate change is a threat to humans and nature. Rise in temperature differs from one region to another around the world. In the general case, warming is relatively higher in land areas than in seas and oceans. Temperature rise is leading to sea level rise, ice melt, ocean precipitation, ocean currents, as well as, major risks to life on land and water. One of the most significant contributors to this threat is greenhouse gas emissions. Understanding the main human factors that cause greenhouse gas emissions is necessary to make data-driven decisions for future actions to combat climate change. In this study, we investigate human activities that contribute to increases in temperature by analyzing new alternative data sources that include internet usage, gas CO2, oil CO2, consumption CO2, air travel, meat consumption, population, car sales, GDP, and housing data, among others. Experimental results indicate that CO2-re1ated features and fertilizer consumption showed a relationship with land temperature. While internet usage patterns did not show any significant correlation, the air travel data showed a reverse effect. The analytics approach and algorithms presented in this paper have the potential to serve as a supplement tool to track and detect emerging predictive features of temperature rise.