This time series study investigated the quarterly banana production among the six provinces in Eastern Visayas, Philippines namely: Biliran, Leyte, Southern Leyte, Samar, Eastern Samar, and Northern Samar from 2010 to 2022 specifically the time series components of the data, appropriate time series model, projected banana production for 2023-2024, and the comparison of the predictive accuracy of forecasted models. The technique employs a descriptive and predictive study design of the secondary data from the Philippine Statistics Authority (PSA) using descriptive statistics, time series charts, Autoregressive Integrated Moving Average (ARIMA) models, forecasting, Mean Absolute Percentage Error (MAPE), and Symmetric Mean Absolute Percentage Error (SMAPE). Among the six provinces, Samar (148,352.78 mt) had the highest total volume of banana production, followed by Southern Leyte (819,306.59 mt). The highest banana production was observed among provinces, namely: 3rd quarter of 2013 in Biliran, 2nd quarter of 2010 in Eastern Samar, 3rd quarter of 2012 in Leyte, 2nd quarter of 2014 in Northern Samar, 1st quarter of 2022 in Samar, and 4th quarter of 2012 in Southern Leyte. In terms of the overall banana production, seasonality was found in quarters from 2010-2022 with irregular variations and gradual increases.
All provinces showed ADF statistics that are negative and p-values that are below the 0.05 threshold, suggesting that the time series for each province is stationary. With ARIMA models being assessed and validated for each province, Eastern Samar (ARIMA(4,1,1)) model has the lowest AIC and BIC values indicating the best fit among the models. Overall ARIMA (3,1,2) model forecasts in Eastern Visayas will experience fluctuations but maintain general stability until 2024. Further, the predictive accuracy using MAE, MAPE, and SMAPE was determined to compare the resulting ARIMA models of the quarterly banana production, hence, the findings revealed variable model accuracies across different provinces with Northern Samar showing the highest accuracy. Thus, the different models and forecasted productions found in this study are important to ensure market stability and consistent supplies of banana production.