Application Research Of The Cloud Model In Time Series Prediction

Time series is a widespread data form of human social activities and objective world, Time series prediction because it provides all kinds of predictions can provide an important basis for the various human activities in all areas of decision-making and has been well received by the worker’s attention. Because of the unique theoretical advantages of combining randomness and fuzziness, cloud model since proposed, has also been a hotspot researchers explore and apply. Therefore, application and research on time series forecasting based on cloud model are discussed in this paper, obviously it’s great significance.Firstly, on the basis of theoretical knowledge of the cloud model is summarized and sorted, the existing theoretical knowledge of cloud model is detailed, and the author’s opinion and interpretation are given. Then, the existing various algorithms of the reverse cloud are summarized and contrasted, a new algorithm that integration advantages of other algorithms is propose. The simulation results revealed that the algorithm has strong robustness and precision compared to other algorithm. Subsequently, based on the analysis of fuzzy theory and existing measurement methods between the cloud models, we proposed two kinds of measurements based on close degree between the cloud models have the same nature concepts: Maximum and Minimum based Cloud Model(MMCM)and Arithmetic Mean and Minimum based Cloud Model(AMMCM).Algorithms are simple especially still maintain the accuracy in few cloud drops. In this paper, except for the new algorithms to do a detailed description, but also for the new approaches with other algorithms gave a comparative analysis of simulation results, finally specific film critics data was applied with. Experimental results showed that the proposed algorithms had a significant optimization on accuracy and algorithm consumption, and had good feasibility and effectiveness in solving classification problems among cloud models.As for application of time series prediction based on cloud model, in this paper, we conducted a study on one of the prediction mechanism, and on this basis, some of algorithms were improved or optimized or provided an alternative method of thinking. First, we improved the method about solving numerical characteristic of cloud model: entropy in the Cloud transform algorithm based on the peak. Then, in the merge process of cloud models on the concept zooming algorithm, another merger condition was mentioned. Experimental results showed that the proposed merger condition makes the concept jumped more reasonable. Ultimately, through empirical analysis, using the highest and the minimum temperature data of Chengdu city in recent years, time series forecasting techniques based on cloud model mentioned in this article was used to predict. Meanwhile the final predicted results were compared to the AR model in traditional time-series prediction, analysis and evaluation were also given in the paper.