Prediction Method Using ARIMA

​

In order to do time series prediction of crypto-currency value using arima, first of all I read the crypto-currency value and corresponding date from the data base. The predictions are mainly based on the opening value of the crypto-currency. Then the cypto currency value is averaged over a month. For that, first I determined how many months are there and  read the corresponding price and doing the average using the aggregate() function of R. Then the data is transferred in time series using the ts() function. Then time series fitting is done in exponential smoothing model of arima using the arima() function. Then the accuracy is being checked using the accuracy() function and the forecast is being predicted using the forecast() function. The error values from arima prediction are as follows:

                                  Mean-Error  Root-Mean-Square-Error   Mean-Absolute-Error  

Bitcoin             79.1906                  1152.301                       419.8816      

Dogecoin        7.195422                   338.8465                      276.8085 

Ethereum        80.78094                   36.0758                      13.0722 

Litecoin        03.041223                   483.8939                      185.5641 

Steller        0.01020263                   14.97211                      4.839247 

BNB        14.89649                   744.6193                      216.8563

Polkadot        3.33119                   144.4437                      82.59183

Solana        0..00360199                   117.6043                      55.15844

​

​

​

​

​

​

​

​