"ANALISIS KOMPARASI MODEL PREDIKSI FINANCIAL DISTRESS SEBAGAI EARLY WARNING SYSTEM PADA PT CENTRAL PROTEINA PRIMA TBK"

Abstrak

Penelitian ini bertujuan untuk menganalisis komparasi tiga model prediksi financial distress, yaitu Springate, Zmijewski, dan Grover, pada PT Central Proteina Prima Tbk periode 2013–2024, serta melakukan forecasting kondisi keuangan periode 2025–2027 berdasarkan model yang terbukti paling akurat. Penelitian menggunakan pendekatan kuantitatif dengan data sekunder berupa laporan keuangan tahunan yang diperoleh dari Bursa Efek Indonesia. Akurasi model dievaluasi melalui kriteria ganda, yaitu tingkat akurasi klasifikasi dan koefisien Cohen's Kappa, dilengkapi metrik Sensitivity, Specificity, Precision, F1-Score, dan Matthews Correlation Coefficient. Perbedaan antarmodel diuji menggunakan Cochran's Q Test, sedangkan proyeksi keuangan dilakukan dengan analisis regresi tren linier melalui estimasi least square. Hasil penelitian menunjukkan bahwa Model Springate dan Zmijewski memiliki akurasi memadai, masing-masing sebesar 75,00% (Kappa 0,5263) dan 91,67% (Kappa 0,8333), sedangkan Model Grover tidak memenuhi kriteria dengan akurasi 66,67% (Kappa 0,3143). Uji Cochran's Q tidak menemukan perbedaan signifikan antarmodel (Q=3,500; p=0,1738). Model Zmijewski terbukti paling akurat dengan skor komposit tertinggi (0,8761). Hasil forecasting memproyeksikan PT Central Proteina Prima Tbk berada dalam kondisi non-distress hingga 2027. Penelitian ini berkontribusi dalam pemilihan model prediksi financial distress yang sesuai untuk sektor akuakultur.

This study aims to compare three financial distress prediction models, namely Springate, Zmijewski, and Grover, applied to PT Central Proteina Prima Tbk during the 2013–2024 period, and to forecast the company's financial condition for 2025–2027 based on the most accurate model. This research employs a quantitative approach using secondary data in the form of annual financial statements obtained from the Indonesia Stock Exchange. Model accuracy was evaluated through dual criteria, namely classification accuracy and Cohen's Kappa coefficient, complemented by Sensitivity, Specificity, Precision, F1-Score, and Matthews Correlation Coefficient metrics. Differences among the models were tested using Cochran's Q Test, while financial projection was conducted through linear trend regression analysis with least square estimation. The results show that the Springate and Zmijewski models possess adequate accuracy, at 75.00% (Kappa 0.5263) and 91.67% (Kappa 0.8333) respectively, whereas the Grover model failed to meet the criteria with an accuracy of 66.67% (Kappa 0.3143). The Cochran's Q Test found no significant difference among the models (Q=3.500; p=0.1738). The Zmijewski model proved to be the most accurate, with the highest composite score (0.8761). The forecasting results project that PT Central Proteina Prima Tbk will remain in a non-distress condition through 2027. This study contributes to the selection of appropriate financial distress prediction models for the aquaculture sector.