Regression analysis on weather phenomena

This thesis explores how four meteorological variables, dew point temperature, air pressure, visibility and total cloud coverage, influence air temperature. Three multiple linear regression models were fitted and compared using diagnostic plots and statistical measures. To test the predictive power of the three models, the last 366 observations, corresponding to the year 1984, were removed. Then, using these observations and the models we examined before, three predictions of the air temperature were made. Overall, this project offers us insight into how some key astronomical factors influence air temperature. By evaluating the predictive accuracy and the statistical measures, we could see the importance some variables have, but also the strength of statistical modeling when it comes to weather forecasts.