The Effectiveness of Autoregression to Predict Temperature

Slight changes in the climate affect people and many aspects of their life. People living in countries around the world, in particular those residing in the underdeveloped countries, begin to feel the effects of climate change. The rapid advancement of statistical method allows the use of certain methods to be used for future prediction. One of the methods is autoregression, a time series type of date, which could effectively be used to predict the weather. Our study is on the effectiveness of autoregression in predicting future temperature. Autoregression is effective at predicting time series data because of the way it looks at past data and trends. When the data is consistent, it will have a high accuracy of prediction. The dataset that was used contained one kind of data, the temperature. Looking at the past data, the trend of the temperature of the areas can be determined. The algorithm took 950 data as a sample data set. Before 285 data were employed for a data test, 655 data were utilized to train the model; the model predicted 285 days. The use of these data was based on the reason that they do not need a large computing power to run the model. The 655 data for training and 285 data for testing the 1/4 data division is enough to create a conclusive model that could predict efficiently. The finding is that autoregression is quite accurate at predicting the temperature with a Mean Absolute Error of 1.4, which means that the model is off by at least 1.4 degree Celcius. In short, autoregression as an algorithm which is used to predict the future by looking at past data is a good way to easily determine the future.