Text2Price: Deep Learning for Price Prediction

Year 2022, Volume: 2 Issue: 2, 28 - 38, 01.10.2022

Beyzanur Saraçlar , Birol Kuyumcu , Selman Delil

Abstract

There are many methods and strategies that can be used when determining the selling price of a product in the online marketplace. Correct pricing of a product is an important factor affecting the overall success and profitability of the e-commerce business. Considering all these issues, the need to develop tools that will help the seller in the process of deciding the price of a product arises. In this paper, we designed a model that predicts the price of a product using its title, supplier, category and description information. Our technique is based on using only a single text data for price estimation. For this purpose, we concatenate product information in a string while preserving their attribute information. The task of preprocessing various feature types becomes simple and quick using this method. The main contribution of our approach is designing a model that is applicable for various prediction tasks without task-oriented implementation. To build the prediction model, we used deep learning methods which are based on RNN and CNN and we compared their performances. According to the results, LSTM-based models have achieved more accurate predictions with 6.1646 mean absolute percentage error (MAPE). Also, CNN-based models had 3x times faster running time advantage while having a minor increase in MAPE with 7.1387 compared to LSTM-based models.

Keywords

price prediction, deep learning, LSTM, CNN

References

• [1] Fathalla, Ahmed, et al. "Deep end-to-end learning for price prediction of second-hand items." Knowledge and Information Systems 62.12 (2020): 4541-4568.
• [2] Carta, Salvatore, et al. "Forecasting e-commerce products prices by combining an autoregressive integrated moving average (ARIMA) model and google trends data." Future Internet 11.1 (2018): 5.
• [3] Tseng, Kuo-Kun, et al. "Price prediction of e-commerce products through Internet sentiment analysis." Electronic commerce research 18.1 (2018): 65-88.
• [4] Kalaiselvi N, Aravind K, Balaguru S, Vijayaragul V (2017) Retail price analytics using backpropogation neural network and sentimental analysis. In: 2017 fourth international conference on signal processing, communication and networking (ICSCN). IEEE, pp 1–6
• [5] Chollet, F., & others. (2015). Keras. GitHub. Retrieved from https://github.com/fchollet/keras
• [6] Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780.
• [7] Graves, A., & Schmidhuber, J. (2005). Framewise phoneme classification with bidirectional LSTM and other neural network architectures. Neural networks, 18(5-6), 602-610.
• [8] E. Grave, A. Joulin, and N. Usunier, “Improving neural language models with a continuous cache,” arXiv preprint arXiv:1612.04426, 2016
• [9] A. Ge ́ron, Hands-on machine learning with Scikit-Learn and Tensor- Flow: concepts, tools, and techniques to build intelligent systems.” O’Reilly Media, Inc.”, 2017.
• [10] Yin, Wenpeng, et al. "Comparative study of CNN and RNN for natural language processing." arXiv preprint arXiv:1702.01923 (2017). [12] Kılınç, D. (2019). A spark‐based big data analysis framework for real‐time sentiment prediction on streaming data. Software: Practice and Experience, 49(9), 1352-1364.
• [11] S. Delil, B. Kuyumcu, C. Aksakallı and İ. S. Akçıra, "Parsing Address Texts with Deep Learning Method," 2020 28th Signal Processing and Communications Applications Conference (SIU), 2020, pp. 1-4, doi: 10.1109/SIU49456.2020.9302154.
• [12] Pedregosa et al., Scikit-learn: Machine Learning in Python, JMLR 12, pp. 2825-2830, 2011.
• [13] Tieleman, Tijmen, and Geoffrey Hinton. "Lecture 6.5-rmsprop: Divide the gradient by a running average of its recent magnitude." COURSERA: Neural networks for machine learning 4.2 (2012): 26-31.
• [14] Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R (2014) Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res 15(1):1929–1958
• [15] Kingma DP, Ba J (2014) Adam: a method for stochastic optimization. arXiv preprint arXiv:1412.6980