What Clothes will be popular? Analysis on MUSINSA item using Embedding and Transfer Learning (fast.ai)
This is machine learning project which analize fashion item to predict cloth's popularity using fast.ai
, especially Collaborative Filtering
and Transfer Learning based Convolutional Neural Network
Data was collected from MUSINSA, a korean fashion web store
The analysis was tested on Google Colab, which is awesome enviromnet for Data Science because most of python package for ds already prepared.
- Python 3.6+
- pytorch
- fast.ai
- BeautifulSoup
- Selenium
- Requests
doc
folder have some contents of Jupter Notebook
- Part0: Data Crawling
- Part1: Data Preprocessing
- Part2: Model Training
- Part3: Model Evaluating
- Part4: Model Serving
- Developing Web Crawling Supporting Module
- Developing fast.ai Supporting Module
- Etc...