Research and Development of Machine Learning Model for Clothing Selection
Students Name: Mostova Mariia Volodymyrivna
Qualification Level: magister
Speciality: Computer Control Systems for Moving Objects (Automobile Transport)
Institute: Institute of Computer Technologies, Automation and Metrology
Mode of Study: full
Academic Year: 2023-2024 н.р.
Language of Defence: англійська
Abstract: A new machine learning model for clothing selection is proposed. Its task is to solve the problem of choosing clothes in any weather. That is, the input data of the system are photos of clothes and weather parameters. The result of this application is a weather-based system based on a cellular neural network. The input data is a photo and a CSV file from a dataset. Before training the model, they were processed and prepared accordingly. The output data after processing and model development is the correlation of clothes to weather conditions. The result of the implemented machine learning model for clothing selection is the creation of new relationships between clothing photos and weather conditions by using the season parameter. That is, the results of this work can be integrated into any system for display in the format of a user interface. In the first section, the reviewed methods and approaches show that research and projects related to weather-based clothing selection are ongoing. Some systems take into account the user’s style, others take into account weather conditions, and some combine both aspects to create more accurate recommendations [1-3]. The second section describes the methods and tools chosen to develop the machine learning model for clothing selection. We chose the Python programming language, TensorFlow, Scikit-learn, and Keras frameworks, Pandas libraries for data manipulation and analysis, and NumPy for numerical computing [4-5]. The third section focuses on preparation before implementing the model, we collected and pre-processed the input data. We chose to build a machine learning model based on a cellular neural network (CNN). The fourth section presents the practical implementation of the machine learning model for clothing selection in the form of Python program code, which consists of training and testing. The fifth section is the economic feasibility study. The obtained results of calculations from the cost price, contract price and planned profit and the high values of the calculated coefficients of scientific and scientific and technical efficiency indicate the economic feasibility of the task. The result of the work is the accuracy of the results of the trained model - 70%, which increases with the number of data and epochs, and the testing - 63%. The developed model classifies test data better than the data it was trained on. Keywords: machine learning model, clothing selection, cellular neural model, weather-based system, training, testing, Python. References: 1. L. Chua, T. Roska “Cellular Neural Networks and Visual Computing: Foundations and Applications”, 2005. 2. Charu C. Aggarwal “Recommender Systems: The Textbook”, 2016. 3. Yujie Liu, Yongbiao Gao, Shihe Feng, Zongmin Li “Weather-to-garment: Weather-oriented clothing recommendation”, IEEE International Conference on Multimedia and Expo, 2017. 4. Wes McKinney “Python for Data Analysis: Data Wrangling with pandas, NumPy, and Jupyter”, 2022, 579 p. 5. Francois Chollet “Deep Learning with Python”, 2017, 384 p.