Prediction Method Using SVM

ISupport vector machine (SVM) is a mainly a classification algorithm. In order to time series prediction a variation of SVM is used which is called the support vector regression (SVR). The major difference between linear regression and SVR is that linear regression mainly tries to minimize the sum of the square error. Linear regression tries to fit a straight line with the data points, calculates the difference between the actual data points and fitted line. This is the fitting error and linear regression tries to minimize the sum of the square error. Whether the SVR algorithm tries to define the hyperplane in such a way so that L2-norm of the coefficient vector gets minimized. In SVR, a margin of error is defined by a slack variable.

​

                                  Training Accuracy     

                    Bitcoin              57.54                      

Dogecoin             62.45

Ethereum             37.56

Litecoin            46.73

Steller            48.63

BNB             39.54

Polkadot.        33.92

Solana.          38.65