ARIMA is a forecasting model that captures historical patterns in time series data but does not incorporate exogenous variables, whereas multiple linear regression includes exogenous variables but does not account for time series dynamics. Therefore, the Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX) model is required to integrate historical data patterns and exogenous influences within a single modeling framework. This study aims to forecast the daily closing price of PT Bank Central Asia Tbk (BBCA) stock using the ARIMAX model with the daily opening value of the Indonesia Composite Index (IHSG) as the exogenous variable, while developing a supporting KNIME workflow. Daily secondary data from September 1, 2023, to September 1, 2025, were analyzed using the Box–Jenkins approach implemented in the open source KNIME software integrated with R. The modeling stages included model identification, parameter estimation using Maximum Likelihood Estimation (MLE), model selection based on the Akaike Information Criterion (AIC), parameter significance testing and residual diagnostics, final model selection based on the principle of parsimony, and visual workflow construction. The results show that the best model is $\text{ARIMAX}(1,1,1,1)$, with a \mbox{Mean Absolute Percentage Error (MAPE)} of $1.089\%$ for in-sample data and $1.213\%$ for out-sample data, indicating highly accurate forecasting performance. Furthermore, the developed KNIME workflow successfully integrated and automated all modeling stages, with 60\% of stages using internal KNIME nodes and 40\% using R Snippet nodes, thereby enabling all modeling stages to be executed automatically within a single integrated pipeline with minimal manual coding.
Copyrights © 2026