Multi Parameter Prediction Using Vector Autoregressive and Autoregressive Integrated Moving Average with Exogenous: A Case Study of Greenhouse of Melon Cultivation in Blitar Regency

Melon cultivation is one of the contributors to economic growth in the Food Crops and Horticulture sub-sector in East Java. Currently, in Wates District, Blitar Regency, there are more than 300 greenhouse units with an average area of 700 $\mathrm{m}^{2}$ for the cultivation of sakata and inthanon melons. Real-time monitoring is needed by farmers to know for certain the quality of greenhouse conditions, especially the complicated conditions of melon cultivation. This study aims to predict real-time multi parameter greenhouse conditions using Vector Autoregressive (VAR). VAR performance compared to the Autoregressive Integrated Moving Average with Exogenous (ARIMAX) model based on MAE and MAPE. The results of this study, the VARI(3,1) model with a data train length of 400 has MAE|MAPE humidity, light intensity, and temperature respectively $0.618|0.655\%$, $2.135|2.735\%$, and $0.079|0.223\%$. VARI(3,1) as a simultaneous model is effectively applied to predict humidity and temperature with an average MAE|MAPE of $0.226|0.277\%$ and $0.029|0.109\%$ respectively with a test data length of 100-6000. The results of light intensity prediction with the VARI(3,1) model have MAE|MAPE is quite large ($1.634|11.124\%$).