Research of Software Solutions to Determine the Optimal Solution for the Specified Parameters

Cover Page

Cite item

Full Text

Open Access Open Access
Restricted Access Access granted
Restricted Access Subscription or Fee Access

Abstract

This paper examines the study of software solutions for optimizing decision-making, in particular, choosing the most appropriate clothing size. The purpose of the work is to conduct research and compare three machine learning methods regarding the issue of predicting clothing size. Software solutions were developed based on an open data set containing user measurements, information about products, sizes and types of ordered goods, reviews and comments on orders. During the work, three machine learning algorithms were implemented: the k-nearest neighbor method, the use of a multilayer fully connected neural network, and the use of a neural network with funny data inputs. Possible solutions and architectures of neural networks are presented and tested regarding the issue of optimizing decision-making regarding size according to the criteria of the user himself. It is proposed to use a neural network with mixed data inputs in the JavaScript programming language using TensorFlow.JS, where mixed inputs mean data on the user’s personal measurements and comments left on the compliance of the declared size. The subsequent implementation of the proposed solution is possible as an independent web application or to integrate the module into web sites with the appropriate subject.

Full Text

Restricted Access

About the authors

Darya B. Pugacheva

MIREA – Russian Technological University

Author for correspondence.
Email: pugacheva@mirea.ru
ORCID iD: 0009-0008-5792-3240

Department of Computer Design, Institute of Advanced Technologies and Industrial Programming

Russian Federation, Moscow

Maya V. Yudina

MIREA – Russian Technological University

Email: yudina_m@mirea.ru

Cand. Sci. (Eng.); associate professor, Department of Computer Design, Institute of Advanced Technologies and Industrial Programming

Russian Federation, Moscow

References

  1. Clothing fit dataset for size recommendation. kaggle. URL: https://www.kaggle.com/datasets/rmisra/clothing-fit-dataset-for-size-recommendation (data of accesses: 21.03.2024).
  2. Results of 2022 from AKIT. E-PEPPER Magazine about E-commerce. URL: https://e-pepper.ru/news/itogi-2022-goda-ot-akit.html?ysclid=lu8p3v8kyj470047044 (data of accesses: 21.03.2024).
  3. Keras: Multiple inputs and mixed data. pyimagesearch. URL: https://pyimagesearch.com/2019/02/04/keras-multiple-inputs-and-mixed-data/ (data of accesses: 21.03.2024).
  4. Kudryavtsev M.A. Methodology of designing a costume model based on a ready-made collection using machine learning. Scientific Research. 2018. No. 6 (26). URL: https://cyberleninka.ru/article/n/metodologiya-proektirovaniya-modeli-kostyuma-na-osnove-gotovoy-kollektsii-s-ispolzovaniem-mashinnogo-obucheniya (data of accesses: 27.03.2024).
  5. Larionova M.A., Babeshko V.N. Prospects for the use of artificial intelligence in light industry. MNIZH. 2021. No. 7-1 (109). URL: https://cyberleninka.ru/article/n/perspektivy-primeneniya-iskusstvennogo-intellekta-v-legkoy-promyshlenosti (data of accesses: 27.03.2024).
  6. Balashov M.K. The future of marketing: neural networks as a customer service tool. Science and Education Today. 2017. No. 11 (22). URL: https://cyberleninka.ru/article/n/buduschee-marketinga-neyronnye-seti-kak-instrument-obsluzhivaniya-klientov (data of accesses: 27.03.2024).
  7. Goodfellow Ya., Benjio I., Courville A. Deep learning. A.A. Slinkin (transl. from English). 2nd ed., cor. Moscow: DMK Press, 2018. 652 p.
  8. Morales M. Grokking deep reinforcement learning. St. Petersburg: Piter, 2023. 464 p. (Series “Programmer’s Library”).
  9. Rishabh M., Mengting Wan, McAuley J. Decomposing fit semantics for product size recommendation in metric spaces. In: Proceedings of the 12th ACM Conference on Recommender Systems. 2018. Pp. 422–426.
  10. Rishabh M., Grover J. Sculpting data for ML: The first act of machine learning. 2021. ISBN: 9798585463570.
  11. Sklearn API Reference. scikit-learn. URL: https://scikit-learn.org/stable/modules/classes.html#module-sklearn.neighbors (data of accesses: 18.03.2024).
  12. Sahoo S., Kumar P., Lakshmi B.R. Mixed data through multiple input for price prediction with multilayer perception and mini VGG. International Journal of Recent Technology and Engineering (IJRTE). No. 8 (2). Pp. 6317–6320.
  13. Ramé A., Sun R., Cord M. MixMo: Mixing multiple inputs for multiple outputs via deep subnetworks. 2020. URL: https://www.researchgate.net/publication/349963564_MixMo_Mixing_Multiple_Inputs_for_Multiple_Outputs_via_Deep_Subnetworks
  14. Linjun Zhang, Zhun Deng, Kawaguchi K. et al. How does mixup help with robustness and generalization? ICLR. 2021.
  15. Borugadda P. et al. Transfer learning VGG16 model for classification of tomato plant leaf diseases: A novel approach for multi-level dimensional reduction. Pertanika Journal of Science and Technology. 2023. N. pag.
  16. Analysis of the clothing market in Russia – demo version of the BusinesStat report. BusinesStat. URL: https://businesstat.ru/images/demo/clothes_russia_demo_businesstat.pdf (data of accesses: 25.03.2024).
  17. The Internet trade market in Russia. Association of Internet Trade Companies – AKIT. URL: https://akit.ru / (data of accesses: 21.03.2024).
  18. TensorFlow models.JS. TensorFlow. URL: https://www.tensorflow.org/js/models?hl=ru (data of accesses: 18.03.2024).

Supplementary files

Supplementary Files
Action
1. JATS XML
Download
2. Fig. 1. Visualization of measurements and ordered sizes for the first 100 users (centimeters, Russian size chart)

Download (238KB)
Indexing metadata
3. Fig. 2. Nearest neighbor parameters for one of the pins

Download (77KB)
Indexing metadata
4. Fig. 3. Neural network architecture for handwritten digit recognition

Download (578KB)
Indexing metadata
5. Fig. 4. Architecture of a fully connected neural network for predicting the size of a set of clothes by three measures

View 
Indexing metadata
6. Fig. 5. Neural network architecture with mixed data inputs used to predict the value of a house

Download (100KB)
Indexing metadata
7. Fig. 6. Neural network architecture with mixed data inputs for predicting the size of a set of clothes

Download (419KB)
Indexing metadata