Following is the prediction made using auto regressive model on australian beer production:
Since in production their is presence of trend and sesonality, lets us check how SARIMAX model behaves, following are the results:
Outliers in ARIMA models disrupt linear relationships and Gaussian error assumptions, leading to biased estimates and inaccurate forecasts. They distort historical patterns, complicating model interpretation and reducing predictive reliability. Refer: ARIMA_Isues_With_Outliers_Or_Spike_Data
Dataset: Walmart Recruiting - Store Sales Forecasting Source: Kaggle