Implementation of the Support Vector Machine (SVM) Algorithm in Predicting Transaction Cancellations at Shopee E-commerce

Pada era digital saat ini, belanja online melalui platform E-commerce seperti Shopee semakin diminati oleh masyarakat. Meskipun Shopee menawarkan kemudahan dan kenyamanan, masalah pembatalan transaksi masih menjadi tantangan utama bagi para penjual, menyebabkan kerugian finansial dan ketidakstabilan operasional. Penelitian ini menggunakan algoritma Support Vector Machine (SVM) untuk memprediksi pembatalan transaksi, yang diharapkan dapat meningkatkan efisiensi dan mengurangi resiko kerugian.

Data yang digunakan dalam penelitian ini berasal dari riwayat transaksi toko nafystore.id di Shopee. Metode yang digunakan adalah CRISP-DM (Cross Industry Standard Process for Data mining), yang meliputi Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, dan Deployment. Data Preparation mencakup proses cleaning dengan metode ‘fillna()’, ‘dropna()’, dan ‘replace’, serta encoding dan normalisasi data sebelum penerapan Principal Component Analysis (PCA) untuk mengurangi dimensi dataset. Teknik resampling Oversampling dengan SMOTE digunakan untuk menangani masalah ketidakseimbangan data.

Model dievaluasi menggunakan K-Fold Cross-Validation dengan tiga, lima, dan sepuluh folds, dan berbagai kernel SVM diuji, termasuk linear, polynomial , radial basis function (RBF), dan sigmoid . Hasil penelitian menunjukkan bahwa model menunjukkan kinerja yang sangat optimal dengan akurasi 95.57%, precision 95.96%%, recall 95.57%%, dan F1-score 95.58%. Confusion Matrix juga digunakan untuk mengevaluasi performa model.

Penelitian ini memberikan kontribusi signifikan dalam mengidentifikasi faktor-faktor yang mempengaruhi pembatalan transaksi di Shopee dan memberikan wawasan penting bagi pengelolaan resiko pembatalan transaksi di platform E-commerce lainnya. Faktor-faktor tersebut yang paling berpengaruh terhadap pembatalan transaksi antara lain Total Pembayaran, Ongkos Kirim yang Dibayar Pembeli, Metode Pembayaran, Voucher yang Ditanggung Penjual, Voucher yang Ditanggung Shopee, Estimasi Pengurangan Ongkos Kirim, Pilihan Pengiriman, dan Waktu Pembayaran Dilakukan.

Penelitian ini memperkaya penggunaan algoritma SVM dalam analisis data E-commerce dan dapat diimplementasikan dalam aplikasi prediksi untuk meningkatkan ketergantungan dan efektivitas transaksi online.

Kata Kunci: E-commerce, pembatalan transaksi, Support Vector Machine (SVM), Shopee, CRISP-DM, Data Preparation, Oversampling, PCA, k-fold Cross-Validation

In today's digital era, online shopping through E-commerce platforms such as Shopee is increasingly in demand by the public. Although Shopee offers convenience and comfort, the problem of transaction cancellation is still a major challenge for sellers, causing financial losses and operational instability. This research uses the Support Vector Machine (SVM) algorithm to predict transaction cancellation, which is expected to improve efficiency and reduce the risk of loss.

The data used in this study comes from the transaction history of the nafystore.id store on Shopee. The method used is CRISP-DM (Cross Industry Standard Process for Data mining), which includes Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Deployment. Data Preparation includes cleaning with 'fillna()', 'dropna()', and 'replace' methods, as well as data encoding and normalization before the application of Principal Component Analysis (PCA) to reduce dataset dimensions. An Oversampling resampling technique with SMOTE was used to handle data imbalance issues.

Models were evaluated using K-Fold Cross-Validation with five folds, and various SVM kernels were tested, including linear, polynomial, radial basis function (RBF), and sigmoid The results showed that the model showed performance with 95.57% accuracy, 95.96% precision, 95.57% recall, and 95.58% F1-score. 95.57%%, and F1-score 95.58%. Confusion Matrix was also used to evaluate the model performance.

This research makes a significant contribution in identifying the factors that influence transaction cancellation in Shopee and provides important insights for managing the risk of transaction cancellation in other E-commerce platforms. The factors that have the most influence on transaction cancellation include Total Payment, Shipping Cost Paid by Buyer, Payment Method, Voucher Covered by Seller, Voucher Covered by Shopee, Estimated Shipping Cost Reduction, Shipping Options, and Time of Payment Made.

This research enriches the use of SVM algorithm in E-commerce data analysis and can be implemented in prediction applications to improve the dependability and effectiveness of online transactions.

Keywords: E-commerce, transaction cancellation, Support Vector Machine (SVM), Shopee, CRISP-DM, Data Preparation, Oversampling, PCA, k-fold Cross-Validation