Perancangan Dashboard Prediksi Data Harga Bahan Baku PT Sukaraja Pangan Utama

Abstract

Informasi merupakan aset penting yang diperoleh dari kemampuan pengelolaan data oleh suatu perusahaan. Perusahaan yang bergerak di bidang industri pangan memiliki berbagai data penting salah satunya data harga bahan baku karena fluktuasinya yang signifikan. PT Sukaraja Pangan Utama merupakan perusahaan yang bergerak di bidang industri pangan membutuhkan dashboard yang dapat memprediksi fluktuasi harga bahan baku di masa depan. Dashboard prediksi dikembangkan dengan metode prototyping yang memiliki lima tahapan yaitu communication, quick plan, modelling quick design, construction of prototype, dan development delivery and feedback. Dashboard prediksi ini memanfaatkan model prediksi hybrid ARIMA dan XGBoost untuk prediksi data harga bahan baku yaitu bawang putih. Dashboard prediksi diuji dengan metode black box testing dengan teknik equivalence partitioning yang seluruh test case berhasil diuji dan mendapat status “Valid”. Model hybrid ARIMA dan XGBoost juga memperoleh nilai uji MAPE sebesar 5,77% yang menunjukkan bahwa model ini cukup akurat. Penelitian ini berhasil mengembangkan dashboard prediksi yang dapat membantu perusahaan dalam mengabil keputusan.

Information is an important asset obtained from a company's data management capabilities. Companies engaged in the food industry have various important data, one of which is raw material price data due to its significant fluctuations. PT Sukaraja Pangan Utama is a company engaged in the food industry that needs a dashboard that can predict future fluctuations in raw material prices. The prediction dashboard was developed using the prototyping method, which consists of five stages those are communication, quick plan, modeling quick design, construction of prototype, and development delivery and feedback. This prediction dashboard utilizes a hybrid ARIMA and XGBoost prediction model for predicting raw material price data, specifically for garlic. The prediction dashboard was tested using the black box testing method with equivalence partitioning techniques, where all test cases were successfully tested and received a “Valid” status. The hybrid ARIMA and XGBoost models also obtained a MAPE test value of 5.77%, indicating that the model is sufficiently accurate. This research successfully developed a prediction dashboard that can assist companies in decision-making.