FADILAH, MUHAMMAD BAGAS (2025) PERBANDINGAN METODE SIMPLE LINEAR REGRESSION, POLYNOMIAL REGRESSION, DAN K-NEAREST NEIGHBORS (KNN) UNTUK PREDIKSI TRANSAKSI PERBANKAN. S1 thesis, Universitas Mercu Buana Jakarta.
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (JURNAL MAHASISWA)
JURNAL-41820110031-MUHAMMAD BAGAS FADILAH.pdf Restricted to Registered users only Download (1MB) |
Abstract
In the digital era, the banking sector faces complex challenges in managingtransaction data and making strategic decisions. Bank XYZ, like many otherbanks, collects a large amount of daily transaction data that includes volume, time, transaction type, and customer profiles. Although this data has great potential to provide valuable insights, it is often not fully utilized to support smarter and more responsive decision-making.This study examines theeffectiveness of three prediction methods in banking transaction analysis: SimpleLinear Regression, Polynomial Regression, and K-Nearest Neighbors (KNN). Transaction data was collected and processed through data cleaning. Subsequently, predictive models were built and tested using training data andtesting data. The prediction results from the three models were compared usingRoot Mean Square Error (RMSE) to determine the best- performing model. Basedon the testing results, Polynomial Regression showed the best performance withan RMSE of 0.1724, while Simple Linear Regression had an RMSE of 0.1844, andK-Nearest Neighbors (KNN) had an RMSE of 0.1739. The conclusion is that, based on the RMSE values of the three methods, Polynomial Regressionoutperformed both Simple Linear Regression and K-Nearest Neighbors (KNN). Keywords: Simple Linear Regression, Polynomial Regression, K-Nearest Neighbors (KNN), Prediction, Data Cleaning, RMSE Dalam era digital, sektor perbankan menghadapi tantangan kompleks dalammengelola data transaksi dan pengambilan keputusan strategis. Bank XYZ, seperti banyak bank lain, mengumpulkan banyak data transaksi harian yang mencakupvolume, waktu, jenis transaksi, dan profil pelanggan. Meskipun data ini memiliki potensi besar untuk memberikan wawasan berharga, seringkali belumdimanfaatkan optimal untuk mendukung pengambilan keputusan yang lebihcerdas dan responsif. Penelitian ini membahas keefektifan tiga metode prediksi dalam analisis transaksi perbankan, yaitu Simple Linear Regression, Polynomial Regression, dan K-Nearest Neighbors (KNN). Data transaksi dikumpulkan dandiproses melalui data cleaning. Setelah itu, model prediksi dibangun dan diuji menggunakan data training dan testing data. Hasil prediksi dari ketiga model dibandingkan menggunakan Root Mean Square Error (RMSE) untuk menentukanmodel dengan performa terbaik. Berdasarkan hasil pengujian ketiga metodemenunujkan bahwa metode Polynomial Regression memiliki hasil yang lebih baikuntuk nilai RMSE 0,1724, sedangkan untuk metode Simple Linear Regressionmendapatkan nilai RMSE 0,1844, dan metode K-Nearest Neighbors (KNN) mendapatkan nilai RMSE 0,1739. Kesimpulan nya adalah berdasarkan nilai RMSE3 metode diatas, Polynomial Regression lebih unggul dibandingkan metodeSimple Linear Regression dan K-Nearest Neighbors (KNN). Kata Kunci: Simple Linear Regression, Polynomial Regression, K-Nearest Neighbors (KNN), Prediksi, Data Cleaning, RMSE
Actions (login required)
 |
View Item |
