Authors : Nabeel Al-A`araji, Eman Al-Shamery and Alyaa Abdul-Hussein
Abstract: Data stream is concerned in industry engineering, finance, economy, traffic and many other fields. The main challenging problems in stream are changed stream with time, time of data arrival and space required for storage stream. Prediction in stream is used to forecast the new data from available data. An Adaptive Regression Neural Network (ARNN) is proposed as a new model in this study based on regression neural network with feedback which it is added to the time variable to make model adaptive for the prediction process with minimum error and high accuracy. The proposed system consists of three main stages, data comes from complex environments may be noisy, redundant, contain outliers and missing values. Thus, the polynomial regression with the segmentation and thresholding are employed for preprocessing stage in the interpolation of missing values and remove outlier points from data stream. The ARNN represents the second and main stage for prediction problem. The evaluation process represents the final stage. The proposed method is compared with traditional regression method for prediction and the results show that the proposed method indicates better accuracy.
Nabeel Al-A`araji, Eman Al-Shamery and Alyaa Abdul-Hussein, 2017. An Optimal Stream Prediction Using Adaptive Regression Neural Network. Journal of Engineering and Applied Sciences, 12: 8844-8850.