Neural network and time series prediction based Research on global warming modelling
Zhiyan Wang, Jianhao Yu · 2023
In order to more accurately study the global warming situation, causes and future trends of global climate change, this paper selects the temperature statistics of two countries in the northern and southern hemispheres, Australia and Afghanistan, in the last hundred years, so as to achieve a more scientific and accurate study of global warming. The research method adopts the neural network model and time series to predict the future temperature change of the two regions until 2100, the innovation is to compare and test the prediction results of different models, and the results show that there are too few feature indicators in the neural network prediction, which leads to the training set and test set deviating from the original data, while the time series relies on the time of each year with its average temperature, which is not much interfered by the external factors, reflecting that the time series can be used for the prediction of global climate change. This shows the advantage of time series in global climate prediction research, then through spearman correlation analysis, it is concluded that the main factor affecting the temperature is geographic location, as well as the conclusion that anthropogenic factors are greater than natural factors, and finally, based on the results of the research, it is proposed to mitigate global warming and put forward the most effective measures.