Skip to content

Latest commit

 

History

History
44 lines (41 loc) · 1.6 KB

README.md

File metadata and controls

44 lines (41 loc) · 1.6 KB
Raw

crypto-viz

A scalable flask webapp to forecast cryptocurrency prices. The forecasting model is built using stacked LSTM for one-to-one sequence of the timeseries.

Model performance:

Cryptocurrency Mean squared error
BTC 1.4925e-04
ETH 6.0898e-05
LTC 5.0990e-05
XRP 2.4237e-04

Requirements

  • binance api
$ python -m pip install python-binance
  • tensorflow 1.x / keras 2.2.5
  • flask
  • numpy
  • pandas
  • matplotlib

Running the webapp

$ cd webapp
$ python3 app.py
Open your internet browser and search for "localhost:5000" in the URL box.

Snapshots

Homepage

homepage

Forecasts for BTC (01/01/2020-01/02/2020)

btc forecast

Forecasts for ETH (01/01/2020-01/02/2020)

eth forecast

Forecasts for LTC (01/01/2020-01/02/2020)

eth forecast

Forecasts for XRP (01/01/2020-01/02/2020)

eth forecast

Realtime deployment

  • add/replace trained models in webapp/cryptoviz/models/
  • use API intergration for extracting prices until current timestamp to predict for future(binance lets you extract prices until 3 hours ago...best case)
  • automate the prediction for continous forecast