Makine Öğrenmesi ve İş Zekası Yöntemleriyle Mobil Uygulamaların Başarısı Arttırılabilir mi?

Year 2020, Issue: 20, 805 - 814, 31.12.2020

Murat Kılınç , Çiğdem Tarhan , Can Aydın

https://doi.org/10.31590/ejosat.793069

Abstract

Son zamanlarda, mobil platformlar için geliştirilen uygulamaların sayısı giderek artmaktadır. Geliştirilen mobil uygulamaların yayınlandığı ana platformlardan birisi olan Google Play Store’da da özellikle açık kaynak kodlu olması sebebiyle, yoğun bir geliştirici ilgisi mevcuttur. Fakat geliştirilen uygulamanın sağlayabileceği başarı ya da hangi özelliklere sahip olması gerektiği gibi unsurlar için geliştiricilerin yararlanabileceği bir platform bulunmamaktadır. Bu çalışmada da bu sorun üzerine gidilmiştir. Bu doğrultuda, geliştirilen uygulamanın özelliklerine göre bir başarı tahminlemesi ve sınıflandırma yapılması amaçlanmıştır. Ayrıca geliştirilen uygulamanın, daha önce geliştirilmiş olan uygulamaların özelliklerine göre iş zekâsı kapsamında değerlendirilmesi de çalışmanın dayanak noktalarından biridir. Araştırma kapsamında, uygulama rating tahminleri için Decision Tree Regressor (DTC), Random Forest Regressor (RFR), K-Neighbors Regressor (KNN) ve AdaBoost Regressor (ABR) kullanılmış ve metriklerin doğruluğu R kare skoru (R2), Mean Square Error (MSE) ve Root Mean Square Error (RMSE) ile test edilmiştir. Sınıflandırma tahminleri için ise Random Forest Classification (RFC), Decision Tree Classification (DTC), K-Neighbors Classification (KNC), MLP Classification (MLP), AdaBoost Classification (ABC) ve Naive Bayes (GNB) algoritmaları kullanılmış ve metriklerin doğruluğu confusion matrix ile test edilmiştir. Bu kapsamda rating tahmini için en iyi sonuçları %80.73 ile DTR ve %82.89 ile RFR, başarı sınıflandırması için en iyi sonuçları ise %86.08 ile DTC, %89.83 ile RFC algoritmaları vermiştir. Çalışma kapsamındaki makine öğrenmesi yönetimleriyle yapılan tüm tahminlemeler dinamik bir şekilde Flask framework kullanılarak web arayüzünde gösterilmiştir. Dolayısıyla, iş zekâsı ile geliştiricilerin karar desteği alabileceği bir platform oluşturulmuş ve ortaya çıkan sonuçlar analiz edilerek çalışma içerisine aktarılmıştır. Bu sayede, mobil uygulama geliştiricileri varsa eksikliklerini görebilecekler ve başarı anlamında bir öngörüye sahip olabileceklerdir.

Keywords

İş Zekası, Makine Öğrenmesi, Web Uygulamaları, Karar Destek Sistemi

Could Mobile Applications' Success be Increased via Machine Learning and Business Intelligence Methods?

Abstract

Recently, the number of applications developed for mobile platforms is increasing. Google Play Store, one of the leading platforms where the developed mobile applications are published, has an intense developer interest due to its open-source code. However, there is no platform that developers can benefit from for the factors such as the success that the developed application can provide or what features it should have. This study also addressed this problem. In this direction, it is aimed to make a success estimation and classification according to the features of the developed application. Also, the evaluation of the developed application within the scope of business intelligence, according to the previously developed applications' characteristics is one of the main points of the study. Within the scope of the research, Decision Tree Regressor (DTC), Random Forest Regressor (RFR), K-Neighbors Regressor (KNN), and AdaBoost Regressor (ABR) were used for application rating estimates and the accuracy of the metrics was determined by the R square score (R2), Mean Square Error (MSE), and Root Mean Square Error (RMSE). Random Forest Classification (RFC), Decision Tree Classification (DTC), K-Neighbors Classification (KNC), MLP Classification (MLP), AdaBoost Classification (ABC) and Naive Bayes (GNB) algorithms were used for classification estimates. The accuracy of the algorithms has been tested with a confusion matrix. In this context, the best results for rating estimation were DTR with 80.73% and RFR with 82.89%, DTC gave the best results for success classification with 86.08% and RFC with 89.83%. All predictions made by machine learning management within the study's scope are dynamically displayed on the web interface using the Flask framework. Therefore, a platform has been created where developers can receive decision support with business intelligence, and the resulting results are analyzed and transferred to the study. In this way, mobile application developers will be able to see their shortcomings and have a prediction in terms of success.

Keywords

Business Intelligence, Machine Learning, Web Applications, Decision Support System

References

• Birant, D. (2011). Comparison of decision tree algorithms for predicting potential air pollutant emissions with data mining models. Journal of Environmental Informatics, 17(1), 46–53. https://doi.org/10.3808/jei.201100186
• Cui, J., Wang, L., Zhao, X., & Zhang, H. (2020). Towards predictive analysis of android vulnerability using statistical codes and machine learning for IoT applications. Computer Communications, 155(February), 125–131. https://doi.org/10.1016/j.comcom.2020.02.078
• Das, T., Di Penta, M., & Malavolta, I. (2020). Characterizing the evolution of statically-detectable performance issues of Android apps. Empirical Software Engineering, 25(4), 2748–2808. https://doi.org/10.1007/s10664-019-09798-3
• Day, M. Y., & Lin, Y. Da. (2017). Deep learning for sentiment analysis on google play consumer review. Proceedings - 2017 IEEE International Conference on Information Reuse and Integration, IRI 2017, 2017-Janua, 382–388. https://doi.org/10.1109/IRI.2017.79
• Dehkordi, M. R., Seifzadeh, H., Beydoun, G., & Nadimi-Shahraki, M. H. (2020). Success prediction of android applications in a novel repository using neural networks. Complex & Intelligent Systems. https://doi.org/10.1007/s40747-020-00154-3
• Erdal, H. (2015). Contribution of Machine Learning Methods to the Construction Industry: Prediction of Compressive Strength. Pamukkale University Journal of Engineering Sciences, 21(3), 109–114. https://doi.org/10.5505/pajes.2014.26121
• García-Peñalvo, F., Conde-González, M., & Matellán, V. (2014). Mobile Apps for Older Users. The Development of a Mobile Apps Repository for Older People. International Conference on Learning and Collaboration Technologies, 117–126. https://doi.org/10.1007/978-3-319-07485-6
• Gentner, D., Stelzer, B., Ramosaj, B., & Brecht, L. (2018). Strategic Foresight of Future B2B Customer Opportunities through Machine Learning. Technology Innovation Management Review, 8(10), 5–17. https://doi.org/10.22215/timreview/1189
• Gupta, L. (2020). Google Play Store Apps Dataset. Web Scraped Data of 10k Play Store Apps for Analyzing the Android Market (Version 6). https://www.kaggle.com/lava18/google-play-store-apps
• Huang, C. J., Liu, M. C., Chu, S. S., & Cheng, C. L. (2005). Application of machine learning techniques to Web-based intelligent learning diagnosis system. Proceedings - HIS’04: 4th International Conference on Hybrid Intelligent Systems, 242–247. https://doi.org/10.1109/ichis.2004.25
• Huber, S., Wiemer, H., Schneider, D., & Ihlenfeldt, S. (2019). DMME: Data mining methodology for engineering applications - A holistic extension to the CRISP-DM model. Procedia CIRP, 79, 403–408. https://doi.org/10.1016/j.procir.2019.02.106
• Lam, M. (2004). Neural network techniques for financial performance prediction: Integrating fundamental and technical analysis. Decision Support Systems, 37(4), 567–581. https://doi.org/10.1016/S0167-9236(03)00088-5
• Lamhaddab, K., Lachgar, M., & Elbaamrani, K. (2019). Porting mobile apps from iOS to Android: A practical experience. Mobile Information Systems, 2019(iii). https://doi.org/10.1155/2019/4324871
• Martín, I., Hernández, J. A., & de los Santos, S. (2019). Machine-Learning based analysis and classification of Android malware signatures. Future Generation Computer Systems, 97, 295–305. https://doi.org/10.1016/j.future.2019.03.006
• Munoz, A., Martin, I., Guzman, A., & Hernandez, J. A. (2015). Android malware detection from Google Play meta-data: Selection of important features. 2015 IEEE Conference on Communications and NetworkSecurity, CNS 2015, 701–702. https://doi.org/10.1109/CNS.2015.7346893
• Olson, D. L., Delen, D., & Meng, Y. (2012). Comparative analysis of data mining methods for bankruptcy prediction. Decision Support Systems, 52(2), 464–473. https://doi.org/10.1016/j.dss.2011.10.007
• Onan, A. (2015). Şirket İflaslarının Tahminlenmesinde Karar Ağacı Algoritmalarının Karşılaştırmalı Başarım Analizi. Bilişim Teknolojileri Dergisi, 8(1), 9–19. https://doi.org/10.17671/btd.36087
• Statcounter. (2020a). Desktop vs Mobile vs Tablet Market Share Worldwide | StatCounter Global Stats. https://gs.statcounter.com/platform-market-share/desktop-mobile-tablet/worldwide
• Statcounter. (2020b). Mobile Operating System Market Share Worldwide | StatCounter Global Stats. https://gs.statcounter.com/os-market-share/mobile/worldwide
• Stoyanov, S. R., Hides, L., Kavanagh, D. J., Zelenko, O., Tjondronegoro, D., & Mani, M. (2015). Mobile App Rating Scale: A New Tool for Assessing the Quality of Health Mobile Apps. JMIR MHealth and UHealth, 3(1), e27. https://doi.org/10.2196/mhealth.3422
• Yang, T. L., Lin, C. H., Chen, W. L., Lin, H. Y., Su, C. S., & Liang, C. K. (2020). Hash Transformation and Machine Learning-Based Decision-Making Classifier Improved the Accuracy Rate of Automated Parkinson’s Disease Screening. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 28(1), 72–82. https://doi.org/10.1109/TNSRE.2019.2950143
• Zhu, H., Cao, H., Chen, E., Xiong, H., & Tian, J. (2012). Exploiting Enriched Contextual Information for Mobile App Classification. CIKM ’12: Proceedings of the 21st ACM International Conference on Information and Knowledge Management, 1617–1621. https://doi.org/10.1145/2396761.2398484