ANALISIS PERBANDINGAN FUNGSI LOSS MSE DAN CAUCHY DALAM PREDIKSI HARGA PENUTUPAN SAHAM NASDAQ

Abstract

This study aims to compare the performance of two loss functions, namely Mean Squared Error (MSE) and Cauchy Loss, in a linear regression model for predicting the closing price of Nasdaq stocks. The data used covers the period from November 15, 2016, to July 15, 2025. Model performance evaluation was conducted using cross-validation and evaluation metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and the coefficient of determination (R²).

The results show that the model with the MSE loss function performs slightly better with an MAE of 79.93 ± 9.06, an MSE of 14,703.26 ± 4,242.03, an RMSE of 120.02, and an R² of 0.9994 ± 0.0002. Meanwhile, the model with the Cauchy loss function recorded an MAE of 82.37 ± 10.56, an MSE of 15,658.23 ± 4,741.97, an RMSE of 123.65, and an R² of 0.9993 ± 0.0002.

Thus, it can be concluded that although both loss functions demonstrate very high predictive performance, the MSE loss function provides more stable and accurate results in the context of predicting Nasdaq stock closing prices using linear regression.

Penelitian ini bertujuan untuk membandingkan performa dua fungsi loss, yaitu Mean Squared Error (MSE) dan Cauchy Loss, dalam model regresi linier untuk memprediksi harga penutupan saham Nasdaq. Data yang digunakan mencakup periode waktu dari 15 November 2016 hingga 15 Juli 2025. Evaluasi performa model dilakukan menggunakan cross-validation dan metrik evaluasi seperti Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), dan koefisien determinasi (R²).

Hasil menunjukkan bahwa model dengan fungsi loss MSE menghasilkan performa yang sedikit lebih baik dengan MAE sebesar 79.93 ± 9.06, MSE sebesar 14,703.26 ± 4,242.03, RMSE sebesar 120.02, dan R² sebesar 0.9994 ± 0.0002. Sementara itu, model dengan fungsi loss Cauchy mencatatkan MAE sebesar 82.37 ± 10.56, MSE sebesar 15,658.23 ± 4,741.97, RMSE sebesar 123.65, dan R² sebesar 0.9993 ± 0.0002.

Dengan demikian, dapat disimpulkan bahwa meskipun kedua fungsi loss menunjukkan performa prediksi yang sangat tinggi, fungsi loss MSE memberikan hasil yang lebih stabil dan akurat dalam konteks prediksi harga penutupan saham Nasdaq menggunakan regresi linier.