Comparison of Time Series Models for Predicting Online Gaming Company Revenue

Öz

Online gaming industry is an area where the effects of any change can be seen in a very short time. Therefore, real-time analysis of revenues, analysis of the commercial performance of the developed content, and rapid monitoring of the revenue contributions of the improvements are essential. Therefore, financial forecasting is a crucial part of business plan which can help strategize how much and how quickly the company intend to grow. In financial forecasting of a given time series, revenue estimations for future will become important research in the industry. This research offers a detailed analysis of recent time series models and focused on both deep learning and statistical methods for time series forecasting on real-world revenue data. Results of the study are examined using one of the leading Finland based online gaming companies’ revenue data. In our experiments, we investigated various time series forecast techniques, such as SARIMA, Theta, Holt Winters, Prophet, Dense Neural Network (DNN), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), N-Beats and Ensemble models. The experimental evaluations illustrate that deep learning models can optimize the financial forecast operations. The result of the study provides insights to managers and analysts in determining the best model to adopt.

Anahtar Kelimeler

• Revenue forecast
• time series analysis;
• SARIMA
• Theta
• Holt-Winters
• Prophet
• DNN
• CNN
• LSTM

Proje Numarası

3211378

Oyun Şirketi Gelirini Tahmin Etmeye Yönelik Zaman Serisi Modellerinin Karşılaştırılması

Öz

Online oyun sektörü, herhangi bir değişikliğin etkilerinin çok kısa sürede görülebildiği bir alandır. Bu nedenle; gelirlerin gerçek zamanlı analizi, geliştirilen içeriğin ticari performansının analizi ve içeriğe yapılan geliştirmelerin sağladığı gelir katkılarının anlık izlenmesi esastır. Bu durum; finansal tahminleri, bir şirkete ne kadar çok ve ne kadar hızlı büyümek istediği konusunda stratejik yardımı olabilecek bir iş planının çok önemli bir parçası haline getirir. Belirli bir zaman serisinin finansal tahmininde geleceğe yönelik gelir tahminleri, endüstride önemli bir araştırma konusu olmaktadır. Bu çalışma, son zamanlardaki zaman serisi modellerinin ayrıntılı bir analizini sunar ve gerçek dünya gelir verileri üzerinde zaman serisi tahmini için hem derin öğrenme hem de istatistiksel öğrenme metodlarına odaklıdır. Araştırmanın sonuçları; Finlandiya merkezli, önde gelen bir çevrimiçi oyun şirketinin gelir verileri kullanılarak incelenmiştir. Deneylerimizde Dönemsel Otoregresif Entegre Hareketli Ortalama (SARIMA/Seasonal AutoRegressive Integrated Moving Average), Theta, Holt Winters, Prophet, Yoğun Sinir Ağları (DNN/Dense Neural Network), Evrişimsel Sinir Ağları (CNN/Convolutional Neural Network), Uzun Kısa Süreli Bellek (LSTM/Long Short-Term Memory), N-Beats ve Ensemble gibi çeşitli zaman serisi tahmin yöntemlerini inceledik. Deneysel değerlendirmeler, derin öğrenme modellerinin, finansal tahmin işlemlerini optimize edebileceğini göstermektedir. Bu çalışmanın sonucu, işletmecilere ve araştırmacılara hangi modeli benimsemenin en iyi olacağına karar verme konusunda farkındalık sağlayacaktır.

Anahtar Kelimeler

• Gelir tahmini
• zaman serisi
• DNN
• SARIMA
• CNN
• Theta
• Holt-Winters
• Prophet

Destekleyen Kurum

TÜBİTAK

Proje Numarası

3211378

Kaynakça

1. [1] G. A. Tularam, T. Saeed, “Oil-price forecasting based on various univariate time-series models” American Journal of Operations Research, 6 (03), 2016, 226.
2. [2] M. S. Wabomba, M. P. Mutwiri and F. Mungai, “Modeling and forecasting Kenyan GDP using autoregressive integrated moving average (ARIMA) models” Science Journal of Applied Mathematics and Statistics, 4(2), 2016, pp 64-73.
3. [3] M. H. Rahman, U. Salma, M. M. Hossain and M. T. F. Khan, “Revenue forecasting using holt–winters exponential smoothing”, Research Reviews: Journal of Statistics, 5(3), 2016, pp 19-25.
4. [4] J. F. Torres, D. Hadjout, A. Sebaa, F. Martínez-Álvarez and A. Troncoso, “Deep Learning for Time Series Forecasting: A Survey”, Big Data, 9 (1), 2021, pp 3-21.
5. [5] S. Siami-Namini, N. Tavakoli and A. S. Namin, “A comparison of ARIMA and LSTM in forecasting time series”, In 17th IEEE International Conference on Machine Learning and Applications (ICMLA), 2018 pp. 1394-1401.
6. [6] C. H. Wu, C. C. Lu, Y. F. Ma and R.S.Lu, "A New Forecasting Framework for Bitcoin Price with LSTM”, IEEE International Conference on Data Mining Workshops (ICDMW), 2018, pp. 168-175.
7. [7] V. Buyar, A. Abdel-Raouf, “A Convolutional Neural Networks-based Model for Sales Prediction”, In Proceedings of the International Conference on Artificial Intelligence, Robotics and Control 2019, pp. 61-67.
8. [8] S.J. Taylor and B. Letham, “Forecasting at scale”, The American Statistician, 72(1), 2018, pp 37-45.

9. [9] E. Zunic, K. Korjenic, K. Hodzic and D.Donko, “Application of facebook’s prophet algorithm for successful sales forecasting based on real-world data” 2018.
10. [10] I. Yenidog˘ an, A. Çayir, O. Kozan, T. Dag˘ and Ç. Arslan, “Bitcoin forecasting using ARIMA and prophet”, In 2018 3rd International Conference on Computer Science and Engineering, (UBMK), 2018, pp. 621-624.
11. [11] T. W. Yoo and I. S. Oh, “Time Series Forecasting of Agricultural Products’ Sales Volumes Based on Seasonal Long Short-Term Memory”, Applied Sciences, 10 (22), 2020, 8169.
12. [12] L. Qian, Y. Fu, and T. Liu, "An Efficient Model Compression Method for CNN Based Object Detection," 2018 IEEE 9th International Conference on Software Engineering and Service Science (ICSESS), 2018, pp. 766-769.
13. [13] S. Chen, and H. He, “Stock prediction using convolutional neural network”, In IOP Conference series: materials science and engineering 435 (1), IOP Publishing, 2018, p. 012026.
14. [14] B. N. Oreshkin, D. Carpov, N. Chapados and Y. Bengio, “N-BEATS: Neural basis expansion analysis for interpretable time series forecasting”, arXiv preprint arXiv:1905.10437, 2018.
15. [15] B. N. Oreshkin, D. Carpov, N. Chapados and Y. Bengio, “Meta-learning framework with applications to zero-shot time-series forecasting”, arXiv preprint arXiv:2002.02887, 2020.
16. [16] S. Makridakis, E. Spiliotis, and V.Assimakopoulos, “The M5 accuracy competition: Results, findings and conclusions”, Int J Forecast, 2020.
17. [17] V. Buyar and A. Abdel-Raouf, “A Convolutional Neural Networks-based Model for Sales Prediction", In Proceedings of the, International Conference on Artificial Intelligence, Robotics and Control, 2019, pp. 61-67.
18. [18] A. Sbrana, A.L.D. Rossi, and M. C. Naldi, “N-BEATS-RNN: Deep learning for time series forecasting”, In 19th IEEE International Conference on Machine Learning and Applications (ICMLA), 2020, pp. 765-768.
19. [19] Global Gaming Market - Growth, Trends, COVID-19 Impact, and Forecasts (2021 - 2026)
20. [20] IAB internet advertising revenue report. Available online: https://www.iab.com/wp-content/uploads/2019/05/Full-Year-2018-IAB-Internet-Advertising-Revenue-Report.pdf (accessed on 10 September 2021)
21. [21] Y. Ju, X. Wang, and X. Chen, "Research on OMR Recognition Based on Convolutional Neural Network Tensorflow Platform", 11th International Conference on Measuring Technology and Mechatronics Automation (ICMTMA), 2019, pp. 688-691.
22. [22] A. Samraj, D. Sovmiya, K. A. Deepthisri and R. Oviya, "Food Genre Classification from Food Images by Deep Neural Network with Tensorflow and Keras", Seventh International Conference on Information Technology Trends (ITT), 2020, pp. 228-231.
23. [23] A. Kolte, B. Mahitha and N.V.G. Raju, "Stratification of Parkinson Disease using python scikit-learn ML library," International Conference on Emerging Trends in Science and Engineering (ICESE), 2019, pp. 1-4.
24. [24] S. Van der Walt, S. C. Colbert, and G. Varoquaux, "The NumPy Array: A Structure for Efficient Numerical Computation," in Computing in Science Engineering, vol. 13, no. 2, March-April 2011, pp. 22-30.
25. [25] N. Ari and M. Ustazhanov, Matplotlib in python," 11th International Conference on Electronics, Computer and Computation (ICECCO), 2014, pp. 1-6.
26. [26] S. G. Iyer and A. D. Pawar, "Machine Learning Model for Predicting Price of Processors using Multivariate Linear Regression," International Conference on Smart Systems and Inventive Technology (ICSSIT), 2019, pp. 52-56.
27. [27] S. Shariar and K. M. Azharul Hasan, "GPU Accelerated Indexing for High Order Tensors in Google Colab," IEEE Region 10 Symposium (TENSYMP), 2020, pp. 686-689.
28. [28] L. Wu and Y. Wang, "Modelling DGM(1,1) under the Criterion of the Minimization of Mean Absolute Percentage Error," Second International Symposium on Knowledge Acquisition and Modeling, 2009, pp. 123-126.
29. [29] C. Reyes, T. Hilaire, S. Paul and C. F. Mecklenbräuker, "Evaluation of the root mean square error performance of the PAST-Consensus algorithm," 2010 International ITG Workshop on Smart Antennas (WSA), 2010, pp. 156-160.
30. [30] F. H. Fard and B. H. Far, "Detecting a certain kind of emergent behavior in multi agent systems applied on mase methodology," 26th IEEE Canadian Conference on Electrical and Computer Engineering (CCECE), 2013, pp. 1-4.