Applied Time Series and Duluth Temperature Prediction

Abstract

Autoregressive integrated moving average (ARIMA) has been one of the popular linear models in time series forecasting during the past three decades.The Triple Expo- nential Model also can be used to fit the time series data. This project takes Duluth temperature predictions as a case study, finding the best statistical model to predict the temperature. I collected 30 years of Duluth monthly maximum temperature data, from 1986 to 2016, and I fi t 29 years of them into di erent models including Triple Exponential Smoothing model, ARIMA model, and SARIMA model. Then I predicted the last year's temperature in those models, and I compared them to the true value of last year's temperature, which gave me the SSE value for each model so that I could find the best model.