Multiple Linear Regression for Domestic Flight Ticket Price Prediction: A Case Study on the Surabaya–Jakarta Route

Regresi Linear Berganda untuk Prediksi Harga Tiket Penerbangan Domestik: Studi pada Rute Surabaya–Jakarta

Abstract

Air travel is valued for its speed and comfort, especially on the busy Surabaya–Jakarta route. However, fluctuating ticket prices complicate planning and challenge airline pricing strategies. This study develops a multiple linear regression model to predict ticket prices using 10,000 observations processed in Google Colaboratory. The workflow includes data preprocessing, transformation, train–test splitting, and assumption testing. Log transformation and robust standard errors address normality and heteroskedasticity, while Ridge Regression mitigates multicollinearity. The final model achieves an R-squared of 96.4%, indicating high accuracy and stability without overfitting. Influential predictors include airline, departure time, flight duration, baggage allowance, and service class, though not all are statistically significant. The model offers practical insights for travelers seeking affordable options and for airlines developing more competitive pricing strategies. It also provides useful references for policymakers and industry stakeholders in promoting efficient and data-driven decisions across the aviation sector.