Evaluating the Trade-Off Between Predictive Performance and Computational Efficiency: The Effect of Stemming on TF-IDF-Based IMDb Sentiment Analysis

Aa Kurniawan Kurniawan; Nurfiqih

doi:10.30998/faktorexacta.v19i2.4606

Authors

Aa Kurniawan Kurniawan Universitas Pamulang Author
Nurfiqih Author

DOI:

https://doi.org/10.30998/faktorexacta.v19i2.4606

Keywords:

IMDb, Sentiment Analysis, Stemming, TF-IDF

Abstract

Sentiment analysis has become an important approach in extracting opinions from textual data. One preprocessing technique frequently used is stemming, which aims to reduce words into their root forms. However, the impact of stemming on classification performance, feature complexity, and computational efficiency remains an open issue. This study evaluates the effect of stemming on IMDb sentiment analysis using four machine learning algorithms: Naïve Bayes, Support Vector Machine (SVM), Decision Tree, and Random Forest. The dataset used consists of 50,000 IMDb movie reviews represented using TF-IDF feature extraction. Evaluation was conducted using accuracy, precision, recall, F1-score, feature complexity analysis, and computational runtime measurement. The results show that stemming successfully reduced TF-IDF features from 99,249 to 68,991 features, representing a feature reduction of 30.49%. However, stemming decreased classification performance across all evaluated algorithms. SVM achieved the highest accuracy of 89.74% without stemming and 89.32% with stemming. Despite the decrease in accuracy, stemming improved computational efficiency, particularly for Decision Tree and Random Forest. These findings indicate that stemming does not always improve sentiment classification performance but can significantly reduce feature complexity and computational cost.

Downloads

Download data is not yet available.

References

[1] A. Utama dan I. Shopi, “Analisis Sentimen Ulasan Film pada IMDb Menggunakan Metode Naive Bayes Classifier dan Support Vector Machine,” Jurnal Teknik Informatika (JTI), pp. 115-124, 2021.

[2] R. Ramadhan dan F. Asrori, “Sentiment Analysis of Movie Reviews Using Text Mining Techniques,” Jurnal RESTI (Rekayasa Sistem dan Teknologi), vol. 6, no. 1, pp. 45-52, 2022.

[3] A. Wahid dan D. Purwitasari, “Pengaruh Tahapan Stemming Terhadap Akurasi Klasifikasi Sentimen Berbasis Teks,” Jurnal Nasional Teknik Elektro dan Teknologi Informasi (JNTETI), vol. 10, no. 3, pp. 241-248, 2021.

[4] A. Hidayatullah dan M. Ma'arif, “Penerapan Text Preprocessing dan TF-IDF untuk Klasifikasi Teks,” Jurnal Teknoinfo, vol. 16, no. 1, pp. 88-95, 2022.

[5] K. Nugroho dan D. Kurniawan, “Evaluasi Efisiensi Waktu Komputasi Algoritma Stemming dalam Analisis Sentimen,” Jurnal Teknologi Informasi dan Ilmu Komputer (JTIIK), vol. 9, no. 4, pp. 789-798, 2022.

[6] E. Prasetyo dan E. Utami, “Analisis Perbandingan Algoritma Klasifikasi Tunggal dan Ensemble untuk Data Berdimensi Tinggi,” JUITA: Jurnal Informatika, vol. 10, no. 2, pp. 167-175, 2022.

[7] L. S. dan W. A., “Peningkatkan Akurasi Klasifikasi Sentimen Menggunakan Algoritma Random Forest”.

[8] J. Putra dan I. Sitanggang, “Komparasi Performansi dan Waktu Komputasi Algoritma Machine Learning pada Teks Mining,” Jurnal Ilmu Komputer dan Informasi (JIKI), vol. 15, no. 1, pp. 12-20, 2022.

[9] F. Hasan, “Penerapan Kerangka Kerja Text Mining untuk Analisis Sentimen Berbasis Komparasi Algoritma,” Jurnal RESTI (Rekayasa Sistem dan Teknologi), vol. 5, no. 4, pp. 712-720, 2021.

[10] N. Ginantra dan G. Dantes, “Analisis Komparasi Efek Preprocessing Text terhadap Akurasi Klasifikasi Sentimen Berbahasa Inggris,” Jurnal Nasional Teknik Elektro dan Teknologi Informasi (JNTETI), vol. 11, no. 4, pp. 289-296, 2022.

[11] H. Barus, I. N. Fajri dan Y. Pristyanto, “Sentiment Classification Analysis of Tokopedia Reviews Using TF-IDF, SMOTE, and Traditional Machine Learning Models,” JAIC, vol. 9, no. 5, pp. 2552-2561, 2025.

[12] Asro dan Solihin, “Comparative Evaluation of Preprocessing Techniques in Twitter Sentiment Analysis for Indonesia’s 2024 Regional Elections,” INOVTEK POLBENG-Seri Informatika, vol. 11, no. 1, pp. 150-159, 2026.

[13] Y. A. Wahid, Sanatang dan D. D. Andayani, “Performance Comparison of Svm and Naïve Bayes For Indonesian-Language Sentiment Analysis On Free Fire Reviews Using Tf-Idf And Smote,” JESSI, vol. 6, no. 4, pp. 674-689, 2025.

[14] J. Sutrisno dan J. G. E. Putri, “Analisis Sentimen Pada Aplikasi Livin’ By Mandiri Menggunakan Metode Tf-Idf Dan Naive Bayes,” JIEMS, vol. 17, no. 1, pp. 1-10, 2024.

[15] I. G. B. A. Budaya dan I. K. P. Suniantara, “Comparison of Sentiment Analysis Algorithms with SMOTE Oversampling and TF-IDF Implementation on Google Reviews for Public Health Centers,” Indonesian Journal of Machine Learning and Computer Science, vol. 4, no. 3, pp. 1077-1086, 2024.

[16] A. S. Muliana, D. Lestarini dan S. P. Raflesia, “Analysis of Public Sentiment on Election Results using Naïve Bayes in Social Media X,” SISTEMASI, vol. 13, no. 6, pp. 2467-2478, 2024.