Klasifikasi Interaksi Kampanye di Media Sosial Menggunakan Naïve Bayes Kernel Estimator

Abstract

The development of technology also influences changes in campaign patterns. Campaign activities are part of the process of Election of Regional Heads. The aim of the campaign is to mobilize public participation, which is carried out directly or through social media. Social media becomes a channel for interaction between candidates and their supporters. Interactions that occur during the campaign period can be one indicator of the success of the closeness between voters and candidates. This study aims to get the pattern of campaign interactions that occur on Twitter social media channels. This interaction pattern is classified as a model in measuring the success of campaigns on social media. The research begins with obtaining data through the data retrieval process using the API feature provided by Twitter. Furthermore, pre-processing is carried out before data can be processed in an algorithmic method. This stage is done to improve data quality so as to improve accuracy. Naive Bayes Classifier was chosen because of a simple procedure, then Kernel Estimator (KE) was used to improve performance. The use of naive Bayes Kernel Estimator can improve model performance from 76.74% to 80.14%. Testing models with split percentage methods on several combinations get satisfactory results.

A Decision Guidance for Solving Success Rate Political Campaign Using Distance Weighted kNN in Nassi-Shneiderman Framework

“…The Ganjar Pranowo dataset for this research has been used before using Naïve Bayes [28]. The results obtained show an accuracy of 76.74%.…”

“…The next step, Naïve Bayes Kernel Estimator is used to improve performance, which gets a yield of 80.14%. From the results of previous studies [28], and referring to a study related to the classification of spam detection on Twitter [29], the k-Nearest Neighbour (kNN) shows better results. A study of the 2019 presidential election in Spain showed the use of the kNN was Figure. 2 Pseudo-code of distance weight kNN able to achieve an accuracy of up to 95% [30].…”

Analisis Sentimen Masyarakat Terhadap Penghapusan Honorer Berdasarkan Opini Dari Twitter Menggunakan Naïve Bayes Classifier

“…NBC menggunakan alat matematika yang dikenal sebagai konsep probabilitas untuk mendeteksi barangkali menjadi klasifikasi terbaik dengan memeriksa frekuensi setiap klasifikasi pada data training [26]. Berikut rumus untuk dasar teorema Naïve Baiyes [27].…”

Repeat Order Recommendation Model to Digitalize the Canteen

“…The canteen is a facility that provides food and drinks for the academic community (Solikin, 2019). In Indonesia, most canteen operations are carried out conventionally, where customers come, queue, officers take orders, and receive orders (Nugroho et al, 2019;Purnama et al, 2020). The cafeteria on campus tends to be crowded at lunchtime or during class changes.…”

“…Digital technology is an opportunity to provide alternative solutions to existing problems. The canteen digitization process was carried out within the scope of food ordering research (Nugroho et al, 2019) which is complemented by the method of recommending repeat orders, delivery services (Andri et al, 2019), and digital payments (Gunawan et al, 2019;Syahril & Rikumahu, 2019), with a case study of the Tenggilis campus canteen, University of Surabaya. The recommendation method used is to analyze comments from customers on the menus that had been ordered and combine them with the history of transactions carried out by these customers (Chen et al, 2016;Wu et al, 2022).…”