A Comparative Study for Temperature Prediction by Machine Learning and Deep Learning

Time series prediction has been a hot research topic for decades. Temperature data is typical time series data, so temperature prediction is an important applications. In recent years, Machine Learning has been applied to time series prediction in many fields. Besides, the development of Deep Learning has made many new breakthroughs in these predictions. Generally, Deep Learning is considered to be superior to traditional Machine Learning in temperature prediction. But in fact, the traditional methods can also perform well for short-term forecast in this regard. This research compares the performance of Linear Regression, SVR, CART, KNN and LSTM models in short-term temperature prediction through three evaluation indicators: MAE, MSE and R2-score. The results explicate that the traditional model can also be better than the depth model in short-term temperature prediction.