8000 GitHub - gov-ai/flowt: Forex and stock market predictions using transformer
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

gov-ai/flowt

Repository files navigation

Flowt

Float like a butterfly, sting like a bee

Flexible framework for realtime time series prediction and online learning. Currently focusing on forex and stock markets. More information will be updated soon.

Project is in it's early stage. To contribute, please contact the author directly.

Requirements

  • Pythion3.8
  • pipenv (install using python3.8 -m pip install pipenv)

Setup Project

$ git clone https://github.com/government-ai/flowt
$ cd flowt
$ python3.8 -m pipenv install
$ python3.8 -m pipenv shell

Change to specific branch

$ git checkout feature/feature-1
$ git branch -a
$ git pull origin feature/feature-1

make your changes using any editor or even file manager.

To see the changes made, use any one of these commands - git status (or) git diff (or) git show

After making code changes, run following commands to update github repo's specific branch

$ git add .
$ git commit -m'comments about changes made'
$ git push origin feature/feature-1

Start API

python -m uvicorn flowt.api.main:app --host 0.0.0.0 --port 8088 --reload

Roadmap

In sequential order

  • setup project folder structure
  • setup tests using pytests
  • setup sphinx for docs
  • setup github workflow for scraping
  • build multihead transformer model in tensorflow
  • implement scraper for investing.com
  • scrape data using github actions
  • send scraped data to telegram channel using cron
  • implement data cleaner
  • implement prediction - online predictior, offline predictor
  • connect to fastapi
  • online training
  • offline re-training
  • connect to database
  • build ui in react and d3.js
  • increase accuracy
  • model optimisation
  • scraper optimisation
  • ...

Create Training Data From Scraped Data

$ python3.7 scripts/clean_data.py data/sample_data data/sample_data_cleaned

About

Forex and stock market predictions using transformer

Resources

Stars

Watchers

Forks

Sponsor this project

Packages

No packages published

Contributors 2

  •  
  •  
0