Stream Prediction Model Based on Tendency Correction

摘要

Linear regression model is widely used in data stream prediction processing. In order to eliminate the prediction deviation caused by small data set, curve tendency correction technique is used to increase the prediction accuracy. Firstly the weighted moving method is used to modify the prediction function parameters. This algorithm improves the predicting accuracy, but causes low efficiency of time and space. Based on this algorithm, the exponential smoothing method is proposed. It is proved that this algorithm can reduce the space and time complexity, and also improves the prediction accuracy.