Analisis Data Sales Performance Pada Industri Penerbangan dan Pengembangan Dashboard Forecasting

Abstract

MOHAMMED ZIDANE. Analisis Data Sales Performance pada Industri Penerbangan dan Pengembangan Dashboard Forecasting. Dibimbing oleh INNA NOVIANTY.

Industri penerbangan menghadapi persaingan ketat dan ketidakstabilan ekonomi yang menuntut analisis sales performance yang akurat. Penelitian ini menerapkan metode Long Short-Term Memory (LSTM) berbasis Extract, Transform, dan Load (ETL) untuk forecasting penjualan 2025 pada PT Nieve Aplikasi Mandiri, mencakup lima variabel utama yaitu, NetFare, Budget, Balance to Go, Growth, dan Achievement menggunakan data historis 2014-2024 dari Oracle Database yang diolah melalui SQL Developer dan Power BI. Evaluasi akurasi menggunakan Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), dan RMSE-Observations Standard Deviation Ratio (RSR) menunjukkan LSTM efektif memprediksi keempat variabel dengan MAPE di bawah 10% (sangat baik), serta seluruh variabel menghasilkan RSR di bawah 0,50 (sangat baik), meski akurasi tetap dipengaruhi kualitas data historis. Hasil forecasting diintegrasikan ke dashboard interaktif Power BI untuk mendukung perencanaan produksi, pengelolaan inventori, alokasi sumber daya, dan pengambilan keputusan strategis berbasis data.

Kata kunci: data historis, dashboard interaktif, forecasting, power bi, sales performance.

MOHAMMED ZIDANE. Sales Performance Data Analysis in the Aviation Industry and Development of Forecasting Dashboard. Supervised by INNA NOVIANTY.

The airline industry faces intense competition and economic instability that demand accurate sales performance analysis. This study applies the Long Short-Term Memory (LSTM) method based on Extract, Transform, and Load (ETL) to forecast 2025 sales at PT Nieve Aplikasi Mandiri, covering five key variables NetFare, Budget, Balance to Go, Growth, and Achievement, using historical data from 2014-2024 sourced from an Oracle Database and processed through SQL Developer and Power BI. Accuracy evaluation using Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), and RMSE-Observations Standard Deviation Ratio (RSR) shows that LSTM effectively predicts four variables with MAPE below 10% (very good category), while all five variables produce RSR values below 0,50 (very good category), although accuracy remains influenced by the quality of historical data. The forecasting results are integrated into an interactive Power BI dashboard that supports production planning, inventory management, resource allocation, and data-driven strategic decision-making.

Keywords: forecasting, historical data, interactive dashboard, power bi, sales performance.