Research Of Program Based On IPTV In Trend Prediction

In the field of social sciences and natural science research, a large number of decision-making problems are inseparable from the forecast, from personal life to the whole country; people will predict the feasibility and merits of the plans, in order to choose the best one. As a basis of decision-making, forecast plays a very important role in the selecting of scheme. At the same time, information about things often is not complete; the theory is not universal or perfect. How to understand things is limited to observational data, namely time series, and then only use the existing historical data to build mathematical models, in order to predict the future.As a new generation of emerging industries, with the rapid development of speed, IPTV is affecting and changing the way of people's lifestyle, gradually. According to the authority of the forecast to2012, China's IPTV subscribers will reach16.2million, and will enter an explosive stage of development. Accompanied by growing, the IPTV platform also gradually has accumulated a large amount of valuable program time series data, belong to the short time sequence, and with a long sequence features. Behind these figures, there contains important reference information for decision-making. Combined with the research project of mentor that namely "User behavior analysis in IPTV platform", this paper studied the time series data mining technology and its application in the prediction of IPTV programs.This paper analyzes the scope of application of the domestic and international time-series forecasting methods. In the background of time-series data which existing in the area of IPTV, this paper presents an effective program hotness mining model and associated algorithms. Combined with the two developments of the program experienced early and late,the program hotness mining model also predict hotness from the two stages, achieved to finish prediction. However, in view of the special tridimensional structure of the IPTV program data which contains users, programs, time. So we transfer the tridimensional date into three dimensional, then use different dimensional to realize the prediction of the different processes of the program.Finally, This article verify the validity of the proposed model and algorithm by analyse the related experi ments, and also provide relevant programming decisions made useful explorations for IPTV.