This notebook is followed by Time series Analysis (ARIMA) articles of machinelearningplus by Selva Prabhakaran and timeseriesplus by Subhasree Chatterjee
ARIMA, short for ‘Auto Regressive Integrated Moving Average’ is essentially a regression model that explains a given time series based on its own past values, in order words, its own lags (past-periods) and the lagged forecast errors, so that can be applied to forecast future values.
This document summarizes how to use ARIMA model, why do we use ARIMA?, the assumptions of ARIMA model with hypothesis test, and the algorithm of time series ARIMA model implementing in daily bitcoin price with computed volatility for predicting values of its cryptocurrency in the future.
Fig 1: Predictive ARIMA model in Time series Forecast for BTC prices compared with their rolling volatilities (i.e. fluctuation)