-1

I am working with multiple time series that aim to predict Bitcoin price. I want to use first time as my index, and then variables x1, x2, x3, and x4 with different combinations to see which combination helps better in predicting bitcoin. When I search over the web, we should have either Univariate or Multivariate time series. With my problem, as I want to predict bitcoin prices based on time, and x1-x4, is it still Univariate or Multivariate? And have you guys seen any implementations on the same problem? All the Univariate problems I saw over the web only deal with time and no other variables, whether in ARIMAX or LSTM.

Mostafa Gholami
  • 19
  • 2

1 Answers1

0

A univariate time series contains only one time dependent variable, while a multivariate time series contains more than one variable. You have a multivariate time series, since you have your variables x1, ..., x3. If you had only the Bitcoin price, it would be univariate.

There are multiple examples available for time series forecasting on stocks, commodities and probably also on Bitcoin. But what you have to remember is that Bitcoin prices (like stocks etc.) do not really dependent on time but on a ton of different factors.

You can find an explanation and example here: https://www.relataly.com/stock-market-prediction-using-multivariate-time-series-in-python/1815/

Arne Decker
  • 808
  • 1
  • 3
  • 9