Fashion Recommendation using machine learning
Motivation and Introduction
When #twinning is a social trend, fashion recommendation will be your new friend. As machine learning is stepping stone in each arena, it can be explored in the fashion world too. Recently, we found different algorithms have been implemented to make choices in online shopping platforms like Amazon, Myntra, Flipkart etc. As we all are aware of prevailing COVID pandemic, everyone are bound to do online shopping. For this, e-commerce sites require a robust recommendation system for their users.
Problem Statement
To develop a model for fashion recommendation using machine learning (ML) techniques. The main objective of this project is to provide a better recommendation system to fulfill the desire of the user.
This is our first ML project which we have done as a requirement of the ML course taken during our course work.
Importance of the project
It is a tedious task to provide expert stylist to an individual user. So, here we came up with ML models that will provide instant stylist recommendation to the users according to their choices.
Data analysis
We have taken the Amazon fashion dataset that is available as an open source and can be easily downloaded. The dataset can be viewed as shown in the figure below:
Methodology
The title from the dataset is pre processed and refined. Further, the pre processed data is used in our model. ML algorithm such as TF-IDF, AVERAGE WORD2VEC, WEIGHTED WORD2VEC, IDF and CNN can be used for building a good fashion recommendation model. We have opted for BAG OF words for our model. The workflow of our model can be explained with the help of following flowchart
Results
A visual representation of our model can be seen in the figure below where we can observe that, on the basis of the apparel chosen by the user, 10 closest apparels similar to the input have been recommended.
References
Shatha Jaradat. 2017. Deep cross-domain fashion rec-ommendation. InProceedings of the Eleventh ACMConference on Recommender Systems, pages 407–410.
P Kamali, P Sudha, and SO Akshaya. Outfit recommender system using knn algorithm. Yuncheng Li, Liangliang Cao, Jiang Zhu, and JieboLuo. 2017.
Mining fashion outfit composition using an end-to-end deep learning approach on set data. IEEE Transactions on Multimedia, 19(8):1946–1955.
Cristiana Stan and Irina Mocanu. 2019. An intelligent personalized fashion recommendation system. In2019 22nd International Conference on Control Systems and Computer Science (CSCS), pages 210–215.IEEE.
Qingqing Tu and Le Dong. 2010. An intelligent personalized fashion recommendation system. In2010International Conference on Communications, Circuits and Systems (ICCCAS), pages 479–485. IEEE.
OR Vincent,AS Makinde,OS Salako, and OD Oluwafemi. 2018.A self-adaptive k-means classifier for business incentive in a fashion design environment. Applied computing and informatics,14(1):88–97.
Contributions
This project was created and developed by our group of two members under the guidance of Dr. Tanmoy Chakraborty :
Kainat Yasmeen: Handling the dataset and coding part for preprocessing, analyzing the results for the ML model. Content creation for the blog.
Akanksha Sneh: Handling the dataset and coding part of applying the ML technique. Handling the implementation of Euclidean algorithm.
