PERBANDINGAN METODE KLASIFIKASI K-NEAREST NEIGHBOR DAN DECISION TREE PADA ANALISIS SENTIMEN APLIKASI DANA DI GOOGLE PLAYSTORE
Khofifah Dwi Fany; irwansyah;
No. Panggil: eSTI-2103015201
ABSTRAK
PERBANDINGAN METODE KLASIFIKASI K-NEAREST NEIGHBOR DAN DECISION TREE PADA ANALISIS SENTIMEN APLIKASI DANA DI GOOGLE PLAYSTORE
Khofifah Dwi Fany
Pesatnya perkembangan dompet digital seperti DANA menimbulkan kekhawatiran terhadap kualitas
layanan yang ditawarkan. Salah satu cara untuk mengevaluasi layanan tersebut adalah melalui
analisis ulasan pengguna di Google Playstore. Banyaknya data yang tersedia membuat analisis
manual menjadi tidak efektif. Penelitian ini mengidentifikasi algoritma klasifikasi yang paling
optimal dalam menganalisis sentimen ulasan aplikasi DANA, dengan membandingkan performa
algoritma K-Nearest Neighbor (K-NN) dan Decision Tree. Data yang digunakan sebanyak 723
ulasan dari platform Kaggle, yang terbagi menjadi 578 data latih dan 145 data uji. Ulasan
diklasifikasikan ke dalam tiga kategori sentimen positif, negatif, dan netral. Proses penelitian
meliputi pengumpulan data, filtering, preprocessing (case folding, tokenizing, stopword removal,
filter token by length), pembobotan TF-IDF, penerapan algoritma K-NN dan Decision Tree, serta
evaluasi menggunakan Confusion Matrix. Hasil menunjukkan bahwa algoritma K-NN memberikan
akurasi sebesar 53,10%, presisi 90,32%, dan recall 41,79%, dibandingkan Decision Tree. Meskipun
recall Decision Tree mencapai 100%, algoritma K-NN lebih direkomendasikan karena memberikan
keseimbangan performa yang lebih baik.
Kata kunci: Analisis Sentimen, K-Nearest Neighbor, Decision Tree, DANA, Google Playstore, Klasifikasi
COMPARISON OF K-NEAREST NEIGHBOR AND DECISION TREE CLASSIFICATION METHODS ON SENTIMENT ANALYSIS OF DANA APPLICATIONS ON GOOGLE PLAYSTORE
Khofifah Dwi Fany
The rapid development of digital wallets such as DANA raises concerns about the quality of the
services offered. One way to evaluate the service is through analyzing user reviews on Google
Playstore. The amount of data available makes manual analysis ineffective. This research identifies
the most optimal classification algorithm in analyzing the sentiment of DANA application reviews,
by comparing the performance of the K-Nearest Neighbor (K-NN) and Decision Tree algorithms.
The data used were 723 reviews from the Kaggle platform, which were divided into 578 training
data and 145 test data. Reviews were classified into three categories of positive, negative and
neutral sentiment. The research process includes data collection, filtering, preprocessing (case
folding, tokenizing, stopword removal, filter token by length), TF-IDF weighting, application of K-
NN and Decision Tree algorithms, and evaluation using Confusion Matrix. The results show that
the K-NN algorithm provides 53.10% accuracy, 90.32% precision, and 41.79% recall, compared to
Decision Tree. Although the recall of Decision Tree reaches 100%, the K-NN algorithm is more
recommended because it provides a better balance of performance.
Keywords: Sentiment Analysis, K-Nearest Neighbor, Decision Tree, DANA, Google Playstore, classification
