Analysis of Temperature Variability, Trends and Prediction in the Karachi Region of Pakistan Using ARIMA Models

In this paper, the average monthly temperature of Karachi region, Pakistan, has been modelled. The time period procured dataset is from January 1989 to December 2018. Autoregressive Integrated Moving Average (ARIMA) modelling technique in conjunction with Box–Jenkins approach applied forecast study area. A total 83.33% trained used for construction model, and remaining 16.67% validation model. best-fitted model identified as ARIMA (2, 1, 4), generated on basis minimum values Akaike Information Criterion (AIC) Bayesian (BIC) procedures. accuracy parameters considered are Root Mean Square Error (RMSE) Absolute (MAE). Both show that 98.152% 98.413% accurate, respectively. addition, Conditional Heteroscedasticity-Lagrange Multiplier (ARCH-LM) test conducted check presence heteroscedasticity residuals This shows no present residual series. By means Autocorrelation Function (ACF) Partial (PACF) plots, most appropriate orders determined evaluated. employed investigate series variables’ precise impact scale regional warming scenario. Accordingly, created can help determining future strategies related weather conditions region. From result, it found seems an increasing trend. Such trend potentially upset economic activities coastal area Pakistan.