Time Series-Based Framework for Global Temperature Forecasting Using Fb-Prophet
摘要
The trends of the Earth’s average temperature are fast growing, and this pace is accelerating. The average annual temperature has increased over 1 \(^\circ \) C and it rises 0.07 \(^\circ \) C on average per ten years. It results in the harshest heat waves to melt glaciers more rapidly than they ever have. Temperature change is known to be caused by greenhouse gas emissions such as CO \(_2\) , followed by CH \(_4\) and N \(_2\) O. Despite this, global emissions have increased by 50% since the year 2000, instead of decreasing. Major factors that cause temperature rise are: burning of fossil fuels, increasing industrialization, human population and deforestation. According to UN figures approximately 11 billion people will live on earth in 2100. It increases global temperatures as more fuels will be burn and land requirement would increase and lead to deforestation. Forecasting future trends, on the basis of historical data is necessary so that precautions can be taken accordingly. Reducing global temperatures will benefit the entire world, since global warming is affecting not only humans but also other living beings. This paper contributes to current research on global warming and climate change. In this study we analysed the global temperature rise and climate change over the years. Since the atmosphere trends are complex, nonlinear systems, traditional methods effective and efficient. So, this study focuses on using Fb-Prophet, a machine learning techniques to forecast the future trends to address the above problem.