Complex Time Series Analysis Based on Conditional Random Fields

A fundamental problem with complex time series analysis involves data prediction and repair.However,existing methods are not accurate enough for complex and multidimensional time series data.In this paper,we propose a novel approach,a complex time series predic-tion model,which is based on the conditional randomfield(CRF)and recurrent neural network(RNN).This model can be used as an upper-level predictor in the stacking process or be trained using deep learning methods.Our approach is more accurate than existing methods in some suitable scenarios,as shown in the experimental results.

Prediction of Time Series Data with Low Latitude Features

The main purpose of this paper is to study the key tech-nology for the prediction of time series data.It has a very wide range of applications,such as forecasting sales.Forecasting sales can be said to play an important role in company operations.Whether for saving costs or inventory scheduling,accurate prediction can save unnecessary waste.From this aspect,this paper uses a neural network to achieve the purpose of the prediction.The application of neural networks in prediction has been a long time.However,most of them have not performed much research on the struc-ture and input of neural networks,and it is not easy to process time series data.Usually,there will be many features.However,the features of data in some scenarios are small.In this paper,we determined how to predict through low-latitude features.Atfirst,among all the ways of preprocess-ing data,the paper selects a mathematical method.After that,this paper builds three models in two aspects:the input and the network structure.To improve the accuracy of the results,this paper proposes two means.One is based on the seasonal characteristics of commodities.The other is based on the prediction error,called exponential smoothing.Finally,according to the results of the experiment,we come to some conclusions.

Dimension Reduction Based on Sampling

Dimension reduction provides a powerful means of reducing the number of random variables under consideration.However,there were many similar tuples in large datasets,and before reducing the dimension of the dataset,we removed some similar tuples to retain the main information of the dataset while accelerating the dimension reduc-tion.Accordingly,we propose a dimension reduction technique based on biased sampling,a new procedure that incorporates features of both dimensional reduction and biased sampling to obtain a computationally efficient means of reducing the number of random variables under consid-eration.In this paper,we choose Principal Components Analysis(PCA)as the main dimensional reduction algorithm to study,and we show how this approach works.