In this paper a symbolic time series forecasting technique is presented which uses range linear regression automata model for forecasting. Range automata model is used to convert the numeric time series into sequence of symbols, this sequence is then fed to linear regression automata model which forecasts the next sample of the time series based on the last three samples of the sequence by using the suffix substrings formed by using sliding window technique. The efficiency of the proposed technique is tested by applying it on the time series medical data of ECG signal, which evidences it to be an operational way for time series forecasting. The performance of RLRA for forecasting is investigated by mean of Mean Absolute Error (MAE) and Error Rate of output from Range Linear Regression Automata Model.
Volume 10 | 14-Special Issue
Pages: 1280-1284