MAPPING DIGITAL LOGISTICS GAPS AND READINESS IN THE CHEMICAL SUPPLIER-PALM OIL CLIENT SUPPLY CHAIN IN PT. KTI INDONESIA: A CASE STUDY

Abstract

The Indonesian palm oil industry has widely adopted digital technologies in upstream plantation activities; however, empirical evidence on digital logistics readiness in downstream input supply chains remains limited. This study addresses this gap by examining digital logistics gaps and organizational readiness in the chemical supplier, palm oil mill supply chain through Machine Learning (ML), based demand forecasting. Using PT KTI, Indonesia, as a case study, the research focuses on forecasting monthly demand for EDTA 4Na, a critical chemical input in palm oil mill operations. Guided by the Drivers–Process–Impact (DPI) framework, the study applies supervised regression models (Random Forest Regressor (RFR) and K-Nearest Neighbors (KNN)) and compares their performance against a Naïve baseline representing heuristic-based manual planning. Model performance is evaluated using Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE), which are interpreted as quantitative indicators of digital logistics gaps and digital readiness. The results show that ML-based approaches outperform the Naïve baseline, demonstrating the feasibility of applying predictive analytics using existing operational data. Among the tested models, KNN exhibits superior performance under conditions of limited data availability and batch-based procurement behavior. Nevertheless, persistent forecasting errors reveal underlying gaps related to data continuity, process standardization, and system integration. Rather than positioning ML solely as a predictive tool, this study contributes to the literature by framing ML-based demand forecasting as a diagnostic mechanism for assessing digital logistics readiness in data-constrained supply chain environments, offering practical insights for early-stage digital transformation in the palm oil industry.