/Ultish-Cryptocurrency-Price-Prediction

Primary LanguagePython

Cryptocurrency-Price-Prediction

Models Employed (Currently)

  1. ARIMA Model
  2. Support Vector
  3. Random Forests
  4. Ridge Regressor
  5. Prophet Model by Facebook (Preferred)

Dataset

The dataset is taken from Yahoo Finance, originally hosted by CoinMarketCap.

Interfacing

Final Interfacing is done through Streamlit of the Prophet Model.

Outputs (For a Single Ticker)

  1. Monthwise High-Low Prices
Monthwise Hi_Low

1. Support Vectors

SVR_Crypto1 SVR_Crypto2

2. ARIMA Model

Arima Crypto_1

3. Random Forest (Output via Notebook)

Random Forest

Second Commit

Added code file for Ridge Regression

Added code file for Prophet model. (R2 score = 0.92)

Below is the output of the Prophet model visualization via Streamlit. a1

a2

a3

a5

Final Model

The Facebooks's Prophet Model is chosen to be the preferred model for both, its accuracy and simplicity while implementing.

References

  1. Crypto Finance Dataset From Yahoo

  2. Stock Price Prediction

  3. Time Series Analysis with Facebook's Prophet

  4. How Does the Prophet Model Works?