Time Series Modeling Method Research Articles

Application of time series model and deep learning method in measuring the impact of COVID-19 on agriculture in Hubei, China

Taking the agricultural situation of Hubei Province as the research object, this paper uses the time series ARIMA model, ARMA model and deep learning LSTM neural network model to explore the impact of COVID-19 on the agriculture of Hubei Province. Three main indicators are screened out to measure the development of agricultural economy, that is, gross regional product, gross output value of agriculture, forestry, animal husbandry and fishery, agricultural product production price index. Based on the quarterly data of indicators from 2001 to 2019 from the National Bureau of Statistics, three indicators in Hubei Province in the first quarter and the second quarter of 2020 are predicted by using the deep learning-based time series ARIMA model, ARMA model and LSTM neural network model. By comparing the predicted data with the real data, the impact of COVID-19 is measured on the agricultural situation of Hubei Province. It was found that COVID-19 had a great impact on the agricultural situation of Hubei Province in both of the first and second quarters of 2020, with the impact in the first quarter being greater than that in the second quarter. At the same time, the prediction accuracy of the two methods is compared to find that the time series model is more effective and reliable in predicting the agricultural product price index. The LSTM neural network model with a long and short term memory has a good prediction effect on the regional gross product and the total output value of agriculture, forestry, animal husbandry and fishery.

Implementation of Using HMM-GA In Time Series Data

Abstract Some time series modeling methods have weaknesses, the static and dynamic information can not be consistently combined. Hidden Markov Model provides solutions to these problems. Hidden Markov Model (HMM) is an extension of the Markov chain where the state cannot be observed directly (hidden), but can only be observed through another set of observations. One of the problems in HMM is how to maximizing P(O|λ) where O is an observation and λ is a model parameter consists of transition matrices, emission matrices, and initial opportunity vectors which can be solved by the Baum-Welch algorithm. In practice, the Baum-Welch algorithm produces a model that is not optimal because this algorithm is very dependent on determining the initial parameters. To solve these problems, HMM will be combined with genetic algorithms (Hybrid GA-HMM). In general, based on AIC and BIC value, Hybrid GA-HMM is optimal than HMM.

Two-Step Meta-Learning for Time-Series Forecasting Ensemble

Amounts of historical data collected increase and business intelligence applicability with automatic forecasting of time series are in high demand. While no single time series modeling method is universal to all types of dynamics, forecasting using an ensemble of several methods is often seen as a compromise. Instead of fixing ensemble diversity and size, we propose to predict these aspects adaptively using meta-learning. Meta-learning here considers two separate random forest regression models, built on 390 time-series features, to rank 22 univariate forecasting methods and recommend ensemble size. The forecasting ensemble is consequently formed from methods ranked as the best, and forecasts are pooled using either simple or weighted average (with a weight corresponding to reciprocal rank). The proposed approach was tested on 12561 micro-economic time-series (expanded to 38633 for various forecasting horizons) of M4 competition where meta-learning outperformed Theta and Comb benchmarks by relative forecasting errors for all data types and horizons. Best overall results were achieved by weighted pooling with a symmetric mean absolute percentage error of 9.21% versus 11.05% obtained using the Theta method.

A close look at 2019 novel coronavirus (COVID 19) infections in Turkey using time series analysis & efficiency analysis.

2019 novel coronavirus (COVID 19) infections detected as the first official records of the disease in Wuhan, China, affected almost all countries worldwide, including Turkey. Due to the number of infected cases, Turkey is one of the most affected countries in the world. Thus, an examination of the pandemic data of Turkey is a critical issue to understand the shape of the spread of the virus and its effects. In this study, we have a close look at the data of Turkey in terms of the variables commonly used during the pandemic to set an example for possible future pandemics. Both time series modeling and popular efficiency measurement methods are used to evaluate the data and enrich the results. It is believed that the results and discussions are useful and can contribute to the language of numbers for pandemic researchers working on the elimination of possible future pandemics.

Predicting market movement direction for bitcoin: A comparison of time series modeling methods

Many traders participate in activities known as "day-trading", trading Bitcoin against the dollar bill as the United States Dollar (USD) on very short timeframes to squeeze out profits from small market fluctuations. This paper aims to help traders decide how to best act by creating a model that can predict price movement's direction for the next 5-min time frame. Several machine-learning models have been tested for this Up/Down binary-classification problem. In this paper, we provide a comparison of the state-of-art strategies in predicting the movement direction for bitcoin, including Random Guessing and a Momentum-Based Strategy. The tested models include Autoregressive Integrated Moving Average (ARIMA), Prophet (by Facebook), Random Forest, Random Forest Lagged-Auto-Regression, and Multi-Layer Perceptron (MLP) Neural Networks. The MLP deep neural network has achieved the highest accuracy of 54% compared to other time-series prediction models. Also, in this paper, various data transformation and feature engineering have been applied in the comparison.

Quantifying cumulative effectiveness of green stormwater infrastructure in improving water quality

Stormwater runoff is one of the main sources of pollution in streams and receiving water bodies of major cities. Green Stormwater Infrastructure (GSI) is a set of distributed stormwater best management practices that absorb excess water, filter out sediment and pollutants, and help recharge groundwater.Despite the increasing popularity of GSI as means of stormwater management, our knowledge of their cumulative performance is limited. In this research, we apply an empirical approach to study the effectiveness of GSI in improving the water quality of four major receiving water bodies of Seattle, Washington, at the watershed scale. We use a Bayesian structural time series model and synthetic control method to build counterfactual scenarios of water quality in absence of GSI implementation and estimate the causal impact of GSI on water quality. We use monthly time series data of water quality parameters (water temperature, dissolved oxygen, surface PAR, chlorophyll a, Secchi depth, pH, light transmission, and fecal coliform) in Seattle's urban watersheds from 2004 to 2017. We also use a set of nine control variables to estimate the counterfactual water quality parameters. Our findings show that GSI improve some water quality parameters such as chlorophyll a, light transmission and Secchi depth, but increase water temperature and decrease dissolved oxygen in some water bodies.

A new deep convolutional neural network model for classifying breast cancer histopathological images and the hyperparameter optimisation of the proposed model

Deep learning algorithms have yielded remarkable results in medical diagnosis and image analysis, besides their contribution to improvements in a number of fields such as drug discovery, time-series modelling and optimisation methods. With regard to the analysis of histopathologic breast cancer images, the similarity of those images and the presence of healthy and tumourous tissues in different areas complicate the detection and classification of tumours on whole slide images. An accurate diagnosis in a short time is a need for full treatment in breast cancer. A successful classification on breast cancer histopathological images will overcome the burden on the pathologist and reduce the subjectivity of diagnosis. In this study, we propose a deep convolutional neural network model. The model uses various algorithms (i.e., stochastic gradient descent, Nesterov accelerated gradient, adaptive gradient, RMSprop, AdaDelta and Adam) to compute the initial weight of the network and update the model parameters for faster backpropagation learning. In order to train the model with less hardware in a short time, we used the parallel computing architecture with Cuda-enabled graphics processing unit. The results indicate that the deep convolutional neural network model is an effective classification model with a high performance up to 99.05% accuracy value.

Forecasting trends in the cryptocurrency exchange rate through the machine learning theory

Subject. The study discusses methodological approaches to forecasting trends in the development of the cryptocurrency market (bitcoin). Objectives. The study aims to discover and explain tools and mechanisms for predicting how the cyptocurrency market may evolve in a short run through time series modeling methods and machine learning methods, which are based on artificial neural networks LSTM. Methods. Using Python-based programming methods, we constructed and substantiated a neural network model for the analyzable series describing how the stock exchange rate of bitcoin develops. Results. Matching loss functions, optimizer and parameters for constructing a neural network that predicts the BTC/USD exchange rate for a coming day, we proved its applicability and feasibility, which is confirmed with the lowest number of errors in the test and validation set. Conclusions and Relevance. The findings mainly prove that the above mechanism is feasible for predicting the cryptocurrency market. The mechanism is based on algorithms for constructing LSTM networks. The approach should be used to analyze and evaluate the current and future parameters of the cryptocurrency market development. The tools can be of interest for investors which operate in new markets of e-money.

Study of nonlinear time series and wavelet power spectrum analysis using solar wind parameters and geomagnetic indices

ABSTRACT We investigate the time series of solar wind parameters (interplanetary magnetic field, Bz and solar wind speed, Vx) and geomagnetic indices (disturbance storm time, Dst and auroral electrojet, AE) using wavelet analysis and nonlinear dynamics time series techniques. The data were collected from the Flight Center Space Physics Data Facility (GSFC/SPDF) OMNIWEB interface between 2008 and 2017. Wavelet power spectrum (WPS) analysis assists in breaking down the time series of Bz, Vx, Dst and AE parameters into different scales. It was noted that there is a greater concentration of power between the 512 and 1024 months bands across the Bz, Vx, Dst and AE parameters. We also applied non-linear time series modelling methods to examine the Bz, Vx, Dst and AE parameters. We utilised both the time delay and embedded dimension in computing average mutual information (AMI) and false nearest neighbors (FNN), respectively. The Lyapunov exponent (LE) is used to express the complexity of the nonlinear dynamics based on embedding parameters. The Lyapunov exponents depict positive values which confirm that the complex solar wind parameters and the geomagnetic indices are deterministic chaotic systems. The results show noticeable chaotic characteristics in the Bz, Vx, Dst and AE parameters.

Personalized Vehicle Trajectory Prediction Based on Joint Time-Series Modeling for Connected Vehicles

Motion prediction for the leading vehicle is a critical task for connected autonomous vehicles. It provides a method to model the leading-following vehicle behavior and analysis their interactions. In this study, a joint time-series modeling approach for leading vehicle trajectory prediction considering different driving styles is proposed. The proposed method enables a precise and personalized trajectory prediction for the leading vehicle based on limited inter-vehicle communication signals, such as the vehicle speed and acceleration of the front vehicles. Three different driving styles are first recognized based on an unsupervised clustering algorithm, namely, Gaussian Mixture Model (GMM). The GMM generates a specific driving style for each vehicle based on the speed, acceleration, jerk, time, and space headway features of the leading vehicle. The feature importance of driving style recognition is also evaluated based on the Maximal Information Coefficient (MIC) algorithm. Then, a personalized joint time series modeling (JTSM) method based on the Long Short-Term Memory (LSTM) Recurrent Neural Network model (RNN) is proposed to predict the front vehicle trajectories. The JTSM contains a common LSTM layer and different fully connected regression layers for different driving styles. The proposed method is tested with the Next Generation Simulation (NGSIM) data on the US101, and I-80 highway dataset. The JTSM is tested for making predictions one to five seconds ahead. Results indicate that the proposed personalized JTSM approach shows a significant advantage over the baseline algorithms.

Dynamic State Estimation for the Advanced Brake System of Electric Vehicles by Using Deep Recurrent Neural Networks

Dynamic state estimation is of considerable importance to the system monitoring, advanced control, and energy management of electrified vehicles (EVs). Among the dynamic states of various vehicle systems, the brake pressure is a key state that reflects the braking intent and maneuver of a driver and is highly correlated with the safety and energy performance of an EV. Thus, it is worth formulating a high-precision estimation algorithm for the brake pressure to better identify the braking intent of a driver and further enhance the multiperformance of the EVs. In this article, an integrated time-series model (TSM) based on multivariate deep recurrent neural networks (RNN) with long short-term memory (LSTM) units is developed for the dynamic estimation of the brake pressure of EVs. The naturalistic driving data are collected using a real electric vehicle under standard driving cycle scenarios. The signals of the vehicle and system states are measured using the controller area network (CAN) bus and preprocessed for model training and prediction. Next, a real-time multivariate LSTM-RNN model for brake pressure estimation is constructed based on the integrated speed estimation model. The real-time scheme iteratively estimates the future velocity and integrates this signal with other vehicle states to estimate a precise value of the braking pressure. The proposed integrated TSM approach is compared with several existing baseline methods to demonstrate the advantage of the method. The testing results indicate that the proposed integrated TSM method can achieve a more reliable multistep prediction with a higher accuracy compared to that of the other methods, which demonstrates the feasibility and effectiveness of the proposed approach.

Wind Energy Forecasting: A Comparative Study Between a Stochastic Model (ARIMA) and a Model Based on Neural Network (FFANN)

The scope of this work is to analyze different time series modeling methods, reported in literature, for forecasting production of electrical energy from renewable energy sources such as wind energy. Two different models are considered: a stochastic model (ARIMA) and a model based on neural networks (FFANN or MLP). The studied time series represent the production of electrical energy from wind of a national power system during March 2018.

Fault Detection for Closed-Loop Control Systems Based on Parity Space Transformation

Fault detection for closed-loop control systems is the future development in the field of the fault diagnosis. Since a closed-loop control system is generally very robust to the external disturbances, fault detection has been challenging a hot research area. Traditional data-driven detection methods are not particularly designed for closed-loop control systems and thus can be improved. In this paper, a new fault detection method is proposed, which is based on the parity space for the closed-loop control system. The main principle of our method is to transform the detection residual into the parity space of the original space to restrict false detection or leak detection caused by the estimation of uncertain states. More specifically, the construction of the stable kernel matrix in the parity space is given, and the residual sequence is accumulated to improve the fault-to-noise ratio and thus increase the detection performance. To verify our method, we have conducted a simulation which is based on a numerical simulation model and the Tennessee industrial system respectively. The results show that the proposed method is more feasible and more effective in fault detection for closed-loop control systems compared with the traditional data-driven detection methods, including the time series modeling method and the partial least squares method.

Evaluation of Drought Condition in Arid and Semi- Arid Regions, Using RDI Index

Investigation of drought event has a great importance in the natural resources management and planning water resources management. One strategy to manage drought is to predict drought conditions by probabilistic tools. In this study climate data of 11 synoptic stations in south of Iran during 1980–2014 were used to estimate of seasonal drought based on RDI index. To prediction of drought (from 2015 to 2020) and analysis of changes trend of it, time series model, first-order Markov Chain model and parametric and non- parametric statistical methods were used. Results showed that MA (5), MA (10), AR (12) and AR (15) were the best time series models that fitted in data of all stations. According to results of prediction of drought classes, classes with normal and moderate dry condition had allocated the most frequency of seasonal drought classes from 2015 to 2020 based on time series model and Markov Chain method. Analysis of changes trend of drought classes showed that based on observed data (1980–2014) and predicted data (1980–2020) changes trend of drought classes in all stations had increasing trend based on parametric and non- parametric statistical methods but increasing trend in about 27% of stations include: Bandar Abbas, Bandar Lengeh, Jask and Shiraz had significant level of 5%. Finally result showed that the study area in 2020 compared to 2014 will be drier.

Novel Time Series Modeling Methods for Gyro Random Noise Used in Internet of Things

In the micro-strapdown inertial navigation systems of Internet of Things, modeling and filtering of gyro random noise are a useful approach to reducing sensor error and enhancing navigation accuracy. Time series is a popular choice for the gyro random noise modeling process. This paper contains two parts to research the modeling methods of gyro random noise. In the first part, for the problem that conventional autoregressive (AR) modeling methods for fiber optic gyro random noise require mass data due to the slow convergence speed of parameter estimation, furthermore, these methods do not perform well if there are too few samples (e.g., less than 3000), and a novel AR modeling method based on the adaptive Kalman filter is proposed. The proposed method can work with smaller sample sizes due to its rapid parameter estimation convergence speed; its algorithm also yields very precise results. These advantages make it well-suited to applications that require a fast AR modeling method for gyro random noise. In the second part, an experiment reveals that the random noise characteristic of some micro-electromechanical systems gyros is adapted to the moving average (MA) model. However, from the available literature, research on a gyro random noise modeling method based on an MA model is very rare. Therefore, we discuss and analyze two MA modeling methods for gyro random noise and give the comparisons between their modeling results.

Time series modeling and filtering method of electric power load stochastic noise

Stochastic noises have a great adverse effect on the prediction accuracy of electric power load. Modeling online and filtering real-time can effectively improve measurement accuracy. Firstly, pretreating and inspecting statistically the electric power load data is essential to characterize the stochastic noise of electric power load. Then, set order for the time series model by Akaike information criterion (AIC) rule and acquire model coefficients to establish ARMA (2,1) model. Next, test the applicability of the established model. Finally, Kalman filter is adopted to process the electric power load data. Simulation results of total variance demonstrate that stochastic noise is obviously decreased after Kalman filtering based on ARMA (2,1) model. Besides, variance is reduced by two orders, and every coefficient of stochastic noise is reduced by one order. The filter method based on time series model does reduce stochastic noise of electric power load, and increase measurement accuracy.

Soft Measurement Modeling Based on Chaos Theory for Biochemical Oxygen Demand (BOD)

The precision of soft measurement for biochemical oxygen demand (BOD) is always restricted due to various factors in the wastewater treatment plant (WWTP). To solve this problem, a new soft measurement modeling method based on chaos theory is proposed and is applied to BOD measurement in this paper. Phase space reconstruction (PSR) based on Takens embedding theorem is used to extract more information from the limited datasets of the chaotic system. The WWTP is first testified as a chaotic system by the correlation dimension (D), the largest Lyapunov exponents (λ1), the Kolmogorov entropy (K) of the BOD and other water quality parameters time series. Multivariate chaotic time series modeling method with principal component analysis (PCA) and artificial neural network (ANN) is then adopted to estimate the value of the effluent BOD. Simulation results show that the proposed approach has higher accuracy and better prediction ability than the corresponding modeling approaches not based on chaos theory.

Comparison of autoregressive moving average and state space methods for monthly time series modelling of Labrador and South-East Quebec river flows

Time series data such as monthly stream flows can be modelled using time series methods and then used to simulate or forecast flows for short term planning. Two methods of time series modelling were reviewed and compared: the well-known auto regressive moving average (ARMA) method and the state-space time-series (SSTS) method. ARMA has been used in hydrology to model and simulate flows with good results and is widely accepted for this purpose. SSTS modelling is a more recently developed method that is relatively unused for hydrologic modelling. This paper focuses on modelling the stream flows from basins of different sizes using these two time series modelling methods and comparing the results. Three rivers in Labrador and South-East Quebec were modelled: the Romaine, Ugjoktok and Alexis Rivers. Both models were compared for accuracy of prediction, ease of software use and simplicity of model to determine the preferred time series methodology approach for modelling these rivers. The SSTS was considered very easy to use but model diagnostics were found to require a high level of statistical understanding. Ultimately, the ARMA method was determined to be the better method for the typical engineer to use, considering the diagnostics were simple and the monthly flows could be easily simulated to verify results.

Time series modeling of paleoclimate data

This paper applies time series modeling methods to paleoclimate series for temperature, ice volume, and atmospheric concentrations of CO2 and CH4. These series, inferred from Antarctic ice and ocean cores, are well known to move together in the transitions between glacial and interglacial periods, but the dynamic relationship between the series is open to question. A further unresolved issue is the role of Milankovitch theory, in which the glacial/interglacial cycles are correlated with orbital variations. We perform tests for Granger causality in the context of a vector autoregression model. Previous work with climate series has assumed nonstationarity and adopted a cointegration approach, but in a range of tests, we find no evidence of integrated behavior. We use conventional autoregressive methodology while allowing for conditional heteroscedasticity in the residuals, associated with the transitional periods. Copyright © 2015 John Wiley & Sons, Ltd.

Clustering Empirical Failure Rate Curves for Reliability Prediction Purposes in the Case of Consumer Electronic Products

In this paper, a methodology based on the combination of time series modeling and soft computational methods is presented to model and forecast bathtub‐shaped failure rate data of newly marketed consumer electronics. The time‐dependent functions of historical failure rates are typified by parameters of an analytic model that grabs the most important characteristics of these curves. The proposed approach is also verified by the presentation of an industrial application brought along at an electrical repair service provider company. The prediction capability of the introduced methodology is compared with moving average‐based and exponential smoothing‐based forecasting methods. According to the results of comparison, the presented method can be considered as a viable alternative reliability prediction technique. Copyright © 2015 John Wiley & Sons, Ltd.