Pengaruh Algoritma ADASYN dan SMOTE terhadap Performa Support Vector Machine pada Ketidakseimbangan Dataset Airbnb

Abstract

Traveling activities are increasingly being carried out by people in the world. Some tourist attractions are difficult to reach hotels because some tourist attractions are far from the city center, Airbnb is a platform that provides home or apartment-based rentals. In lodging offers, there are two types of hosts, namely non-super host and super host. The super-host badge is obtained if the innkeeper has a good reputation and meets the requirements. There are advantages to being a super host such as having more visibility, increased earning potential and exclusive rewards. Support Vector Machine (SVM) algorithm classification process by these criteria data. Data set is unbalanced. The super host population is smaller than the non-super host. Overcoming the imbalance, this over sampling technique is carried out using ADASYN and SMOTE. Research goal was to decide the performance of ADASYN and sampling technique, SVM algorithm. Data analyses used over sampling which aims to handle unbalanced data sets, and confusion matrix used for testing Precision, Recall, and F1-SCORE, and Accuracy. Research shows that SMOTE SVM increases the accuracy rate by 1 percent from 80% to 81%, which is influenced by the increase in the True (minority) label test results and a decrease in the False label test results (majority), the SMOTE SVM is better than ADASYN SVM, and SVM without over sampling.

Detection and identification of Red Snapper (Lutjanus gibbus and Lutjanus malabaricus) and grouper (Plectropomus leopardus and Plectropomus maculatus) with deep learning

“…Recall is a value that indicates the level of success or specificity in retrieving information correctly about negative class data or positive text content [15]. F1-Score is a comparison of precision and weighted average gain [16]. In this study, the F1-Score was 93.38%, which means that the designed model is considered good in classifying and predicting fish species and estimating the length of dead fish.…”

“…The Confusion Matrix method is often used to measure performance on the grouping method. This method works by comparing the data that has been tested into four categories of embedded terms to describe the process results, namely False Positive (FP), False Negative (FN), True Positive (TP), and True Negative (TN) [10]. A confusion matrix can be used as a calculation of various performance metrics to measure the performance of a model that has been created.…”

Pemetaan Opini Publik Terhadap Perubahan Kebijakan BPJS Kesehatan Dengan Pendekatan Support Vector Machine(svm) Dalam Analisis Sentimen

“…Proses evaluasi dengan menggunakan crossvalidation dan perhitungan manual confusion matrix melibatkan pengukuran akurasi, presisi, recall, dan F1-Score. [9] Akurasi mencerminkan sejauh mana model dapat mengklasifikasikan dengan tepat.…”

The Performance Comparison of Classification Algorithm in Order to Detecting Heart Disease

“…Metode oversampling pada kelas minority dapat membantu mencapai kinerja klasifikasi yang lebih baik. Pada penelitian yang dilakukan oleh Hidayat et al [9] melakukan perbandingan performa algoritma klasifikasi dari data yang tidak seimbang menggunakan metode SMOTE dan ADASYN. Penerapan metode SMOTE dan ADASYN pada data yang tidak seimbang sangat mempengaruhi performa algoritma SVM, terjadi kenaikan kenaikan pada label True (minoritas) pada pengujian pada F1-Score dari 0.597 menjadi 0.815 begitu juga pada presisi dan recall.…”