With the evolution of the electricity market into a restructured smart version, load forecasting has emerged as an eminent research domain. Many forecasting models have been proposed by researchers for electricity price and load forecasting. This state of art introduces a load time series modeled with a hybrid technique culminating from the logical amalgamation of GARCH, a conventional hard computing method, Fuzzy ARTMAP, an artificial intelligence-based soft computing technique, and wavelet transform, for treating the load time series. The study investigates into the ability of the proposed hybrid model in tackling the electricity load time series forecasting problems. The work under this study also includes comparisons drawn among models which use either one or two of the mentioned techniques and the model proposed. Results certify the efficacy and effectiveness of the model over others.

When using the microservice architecture to build a software development knowledge base, due to the differences in the user's knowledge level, different users have different needs for knowledge, which causes the problem of unpredictable system load. This paper proposes a load prediction model based on User Knowledge Level Evolution (UKLE). We use user historical access data and user private data to build user portraits, learn the evolution law of user knowledge levels, and thus describe the diverse knowledge growth paths of users. Finally, we predict the future knowledge access trends of users. And we use a hybrid prediction model based on linear and non-linear methods to perform load prediction for service load. The final load prediction model combines the characteristics of user knowledge level evolution and a hybrid prediction model based on service historical load, which improves the model's prediction accuracy and robustness.

Trying to predict the future price of certain stocks, securities or indexes is quite a common goal, being motivated by different reasons and being based on various techniques. The present article has the same purpose, employing an ARIMA model, due to its proven effectiveness and success. Used data is comprised of monthly values for the mentioned index, on a four-year period, from 2010 to 2014, which lead to 60 recordings. The main steps for the analysis are identifying the model, estimating the parameters and the prediction itself. After each one of them is carefully conducted, a comparison is made: the predicted values for BET versus the real values for BET, in order to see if any resemblances exists, or if the differences are significant. Those resemblances or differences are explained, while the conclusion will highlight ARIMA’s capacity or incapacity of forecasting in an accurate way, in the presented context.

Today, enterprise applications impose more and more resource requirements to support an ascending number of clients and to deliver them an acceptable Quality of Service (QoS). To ensure such requirements are met, it is essential to apply appropriate resource and application monitoring techniques. Such techniques collect data to enable predictions and actions which can offer better system performance. Typically, system administrators need to consider different data sources, so making the relationship among them by themselves. To address these gaps and considering the context of general networked-based systems, we propose a survey that combines a discussion about system monitoring, data prediction, and resource management procedures in a unified view. The article discusses resource and application monitoring, resource management, and data forecast at both performance and architectural perspectives of enterprise systems. Our idea is to describe consolidated subjects such as monitoring metrics and resource scheduling, together with novel trends, including cloud elasticity and artificial intelligence-based load prediction algorithms. This survey links the aforesaid three pillars, emphasizing relationships among them and also pointing out opportunities and research challenges in the area.

Time series analysis is an essential research area for those who are dealing with scientific and engineering problems. The main objective, in general, is to understand the underlying characteristics of selected time series by using the time as well as the frequency domain analysis. Then one can make a prediction for desired system to forecast ahead from the past observations. Time series modeling, frequency domain and some other descriptive statistical data analyses are the primary subjects of this study: indeed, choosing an appropriate model is at the core of any analysis to make a satisfactory prediction. In this study Fourier and wavelet transform methods are used to analyze the complex structure of a financial time series, particularly, S&P500 daily closing prices and return values. Multiresolution analysis is naturally handled by the help of wavelet transforms in order to pinpoint special characteristics of S&P500 data, like periodicity as well as seasonality. Besides, further case study discussions include the modeling of S&P500 process by invoking linear and nonlinear methods with wavelets to address how multiresolution approach improves fitting and forecasting results.

This study aims to introduce an ideal model for forecasting crude oil price volatility. For this purpose, the ‘predictability’ hypothesis was tested using the variance ratio test, BDS test and the chaos analysis. Structural analyses were also carried out to identify possible non-linear patterns in this series. On this basis, Lyapunov exponents confirmed that the return series of crude oil price is chaotic. Moreover, according to the findings, the rate of return series has the long memory property rejecting the efficient market hypothesis and affirming the fractal markets hypothesis. The results of the Geweke and Porter-Hudak test verified that both the rate of return and volatility series of crude oil price have the long memory property. Besides, according to both mean square error (MSE) and root mean square error (RMSE) criteria, wavelet-decomposed data improve the performance of the model significantly. Therefore, a hybrid model was introduced based on the long memory property which uses wavelet decomposed data as the most relevant model.

This paper proposes the Discrete Wavelet Transform (DWT), Auto Regressive Integrated Moving Averages (ARIMA) model and Recurrent Neural Network (RNN) based technique for forecasting the computer network traffic. Computer network traffic is sampled on computer networking device connected to the internet. At first, discrete wavelet transform is used to decompose the traffic data into non-linear (approximate) and linear (detailed) components. After that, detailed and approximate components are reconstructed using inverse DWT and predictions are made using Auto Regressive Moving Average (ARIMA) and Recurrent Neural Networks (RNN), respectively. Internet traffic is a time series which can be used to predict the future traffic trends in a computer network. Numerous computer network management tasks depend heavily on the information about the network traffic. This forecasting is very useful for numerous applications, such as congestion control, anomaly detection, and bandwidth allocation. Our method is easy to implement and computationally less expensive which can be easily applied at the data centers, improving the quality of service (QoS) and reducing the cost.

This paper proposes an efficient time series fore- casting model for exchange rates. Previous literature reveals that Functional Link Artificial Neural Network (FLANN) is very effective in financial time series forecasting involving less computational load and fast forecasting capability. Autoregressive Integrated Moving Average (ARIMA) models are well known for their remarkable forecasting accuracy. In this literature, we have used Discrete Wavelet Transform (DWT) to decompose the in-sample training data into linear (detailed) and nonlinear (approximate) components, then applied ARIMA and FLANN model to forecast the respective components. The proposed method amalgamate the unique strengths of ARIMA, FLANN and DWT to improve the forecasting accuracy of a financial time series data. Simulation results show superiority of the proposed method.

This article suggests an imperial real world problem technique for forecasting the financial time series data in order to solve real problems. The financial data are forecasted via ARIMA model in order to give some future suggestion about the case study. Some stock market data have been obtained from the department of land and survey in Jordan. Thus, this was aimed at implementing this model.

The research objective is to discuss the adoption of the wavelet transformation method (WT) in processing time series, for its efficiency. As well as comparing modern methods represented by wavelet and neural networks with traditional methods represented by Box-Jenkins models, to better diagnose the treatment of any series, whether (linear, semi-linear or non-linear), a way to minimize the error to the nearest zero value, through the use of error accuracy measurements to diagnose the best method among classical and modern methods, some of them are characterized by artificial intelligence (IT) to measuring the accuracy of best forecasting methods in time series. Artificial neural networks (ANN) also used as one of the uses of artificial intelligence (IT) for best results. statistical error criteria have been adopted for comparing and evaluating the efficiency of the methods adopted: (MSE, RMSE, MAPE), Mat lab 8th edition was used. An important conclusion reached, was finding the best technique that minimizes the error to its lowest value with an average error close to zero significantly.

The practical significances and complexities of financial time series analysis induce highly demand more reliable hybrid model that denoised the data efficiently, handled with both linear and nonlinear patterns in the data, to achieve more accurate results. This paper suggests a new forecasting hybrid model for financial time series data combined Empirical Wavelet Transform (EWT) technique with improved Artificial Bee Colony (ABC) algorithm, Extreme Learning Machine (ELM) neural network, and Auto-Regressive Integrated Moving Average (ARIMA) linear analysis algorithm. The EWT is used to decompose and denoise the data to reconstruct the data more suitable for forecast. The improvement of the ABC algorithm is according to the Good Point Sets (GPS) theory and adaptive Elite-based Opposition (EO) strategy (GPS-EO-ABC) to overcome the drawbacks of the original algorithm and enhance the optimization performance. The optimized ELM with GPS-EO-ABC, as well as the ARIMA, are utilized independently to generated different forecasting results and combined by the weight-based procession. We testify the performance of the proposed improved ABC algorithm by ten benchmark functions, simulating the proposed forecasting models by three financial time-series datasets. The results indicate that: (1) The proposed algorithm shows outstanding capacities in parameter optimization. The optimized ELM generated more stable and precise results compared with original ELM, ABC-ELM, single LSTM, and ANN; (2) The proposed hybrid model has not only effectiveness but also efficiencies in denoising data, correcting outliers, coordinating both linear and nonlinear patterns, its performance in financial time series forecasting is more excellent than existing hybrid models.

The data mining and its tool has played a vital role in exploring the data from different ware houses. Using data mining tools and analytical technologies we do a quantifiable amount of research to explore new approach for the investment decisions .The market with huge volume of investor with good enough knowledge and have a prediction as well as control over their investments. The stock market some time fails to attract new investor. The reason states that non-aware and also people don’t want to come forward to fall in to the risk. An approach with adequate expertise is designed to help investors to ascertain veiled patterns from the historic data that have feasible predictive ability in their investment decisions. In this paper the NSE – Nifty Midcap50 companies among them top 4 companies having max Midcap value has been selected for analysis. The historical data has a significant role in, helping the investing people to get an overview about the market behavior during the past decade. The stock data for the past five years has been collected and trained using ARIMA model with different parameters. The test criterions like Akaike Information Criterion Bayesian Information Criterion (AICBIC) are applied to predict the accuracy of the model. The performance of the trained model is analyzed and it also tested to find the trend and the market behavior for future forecast.

The blast furnace (BF) is the heart of the integrated iron and steel industry and used to produce melted iron as raw material for steel. The BF has very complicated process to be modeled as it depends on multivariable process inputs and disturbances. It is very important to minimize operational costs and reduce material and fuel consumption in order to optimize overall furnace efficiency and stability, and also to improve the lifetime of the furnace within this task. Therefore, if the actual flame temperature value is predicted and controlled properly, then the operators can maintain fuel distribution such as oxygen enrichment, blast moisture, cold blast temperature, cold blast flow, coke to ore ratio, and pulverized coal injection parameters in advance considering the thermal state changes accordingly. In this paper, artificial neural network (ANN), multiple linear regression (MLR), and autoregressive integrated moving average (ARIMA) models are employed to forecast and track furnace flame temperature selecting the most appropriate inputs that affect this process parameter. All data were collected from Erdemir Blast Furnace No. 2, located in Eregli, Turkey, during 3 months of operation and the computational results are satisfactory in terms of the selected performance criteria: regression coefficient and root mean squared error. When the proposed model outputs are considered for the comparison, it is seen that the ANN models show better performance than the MLR and ARIMA models.

Stock price prediction has always attracted interest because of the direct financial benefit and the associated complexity. From our literature review, we felt the need of a study having sector specific analysis with a broad range of stocks. In this paper, we have conducted a study on the effectiveness of Autoregressive Integrated Moving Average (ARIMA)model, on fifty six Indian stocks from different sectors. We have chosen ARIMA model, because of its simplicity and wide acceptability of the model. We also have studied the effect on prediction accuracy based on various possible previous period data taken. The comparison and parameterizat ion of the ARIMA model have been done using Akaike information criterion (AIC). The contribution of the paper , are a) coverage of a good number of Indian stocks b) Analysis of the models based on sectors c) Analysis of prediction accuracy based on the var ying span of previous period data.

Soil temperature (ST) is an important dynamic parameter, whose prediction is a major research topic in various fields including agriculture because ST has a critical role in hydrological processes at the soil surface. In this study, a new linear methodology is proposed based on stochastic methods for modeling daily soil temperature (DST). With this approach, the ST series components are determined to carry out modeling and spectral analysis. The results of this process are compared with two linear methods based on seasonal standardization and seasonal differencing in terms of four DST series. The series used in this study were measured at two stations, Champaign and Springfield, at depths of 10 and 20 cm. The results indicate that in all ST series reviewed, the periodic term is the most robust among all components. According to a comparison of the three methods applied to analyze the various series components, it appears that spectral analysis combined with stochastic methods outperformed the seasonal standardization and seasonal differencing methods. In addition to comparing the proposed methodology with linear methods, the ST modeling results were compared with the two nonlinear methods in two forms: considering hydrological variables (HV) as input variables and DST modeling as a time series. In a previous study at the mentioned sites, Kim and Singh Theor Appl Climatol 118:465-479, (2014) applied the popular Multilayer Perceptron (MLP) neural network and Adaptive Neuro-Fuzzy Inference System (ANFIS) nonlinear methods and considered HV as input variables. The comparison results signify that the relative error projected in estimating DST by the proposed methodology was about 6%, while this value with MLP and ANFIS was over 15%. Moreover, MLP and ANFIS models were employed for DST time series modeling. Due to these models’ relatively inferior performance to the proposed methodology, two hybrid models were implemented: the weights and membership function of MLP and ANFIS (respectively) were optimized with the particle swarm optimization (PSO) algorithm in conjunction with the wavelet transform and nonlinear methods (Wavelet-MLP & Wavelet-ANFIS). A comparison of the proposed methodology with individual and hybrid nonlinear models in predicting DST time series indicates the lowest Akaike Information Criterion (AIC) index value, which considers model simplicity and accuracy simultaneously at different depths and stations. The methodology presented in this study can thus serve as an excellent alternative to complex nonlinear methods that are normally employed to examine DST.

Nowadays, stock market data forecasting has drawn a high attention in the field of nonstationary and nonlinear time series data with a high heteroscedasticity, since improving the forecasting accuracy is a hot topic for the researchers. Therefore, in this article the authors are proposed a new methodology via combining Empirical Mode decomposition and Moving Average model as a modified method to improve forecasting accuracy in content of stock market data. The strength of this proposed methodology lies in its ability to forecast nonlinear and non-stationary financial data without a need to use any transformation method. Moreover, this method provides a better model with sufficient forecasting accuracy. The daily stock market data of fourteen countries is applied to show the forecasting performance of the proposed method. Based on the five forecast accuracy measures, the results indicate that proposed forecasting method performance is superior to four selected forecasting techniques.

Modern machine learning models (such as deep neural networks and boosting decision tree models) have become increasingly popular in financial market prediction, due to their superior capacity to extract complex non-linear patterns. However, since financial datasets have very low signal-to-noise ratio and are non-stationary, complex models are often very prone to overfitting and suffer from instability issues. Moreover, as various machine learning and data mining tools become more widely used in quantitative trading, many trading firms have been producing an increasing number of features (aka factors). Therefore, how to automatically select effective features becomes an imminent problem. To address these issues, we propose DoubleEnsemble, an ensemble framework leveraging learning trajectory based sample reweighting and shuffling based feature selection. Specifically, we identify the key samples based on the training dynamics on each sample and elicit key features based on the ablation impact of each feature via shuffling. Our model is applicable to a wide range of base models, capable of extracting complex patterns, while mitigating the overfitting and instability issues for financial market prediction. We conduct extensive experiments, including price prediction for cryptocurrencies and stock trading, using both DNN and gradient boosting decision tree as base models. Our experiment results demonstrate that DoubleEnsemble achieves a superior performance compared with several baseline methods.

Many researchers documented that the stock market data are nonstationary and nonlinear time series data. In this study, we use EMD-HW bagging method for nonstationary and nonlinear time series forecasting. The EMD-HW bagging method is based on the empirical mode decomposition (EMD), the moving block bootstrap and the Holt-Winter. The stock market time series of six countries are used to compare EMD-HW bagging method. This comparison is based on five forecasting error measurements. The comparison shows that the forecasting results of EMD-HW bagging are more accurate than the forecasting results of the fourteen selected methods.

Many applications have been done in the field of using wavelet analysis for time series analysis. In this study, we used the quarterly data of Electric Energy Supply in Duhok Province-Iraq in Megawatt which represents a sample size (46) observations during the period 2004 and 2015.we aim to describe how wavelet de-noising can be used in time series forecasting and improve the forecasting quality through presenting some proposed methods based on wavelet analysis and SARIMA method and applying on real data and make comparison between methods depending on some statistical criteria.Results from the analysis showed the superiority of the three proposed methods and showed that we can get more information from a series when using Wavelet-SARIMA method and this leads to enhance the classical SARIMA model in forecasting. Furthermore, after many empirical experiments with many wavelet families, it has been found that Daubechies, Coiflets, Discrete Meyer(dmey) and Symlet wavelets are very suitable when denoising the data and out of these four wavelet families, the Daubechies and Discrete Meyer performed better.

In this paper we propose a novel approach for time series forecasting based on ordered weighted averaging operators (OWA) as linear filter and forecasting models. The OWA operators describe a family of averaging operators parameterized by the choice the weights or filter coefficients. Starting with a nonstationary time series of a given phenomenon, we evaluate the use of linear decaying and constant weights as filtering processes of time series. Moreover, we investigate the effectiveness of using exponential weighted moving average process as a filter linear. After the application of the linear operators, we formulate the best possible forecasting models, ARIMA and neural network models, for short-term forecasting for each of the new structured time series using the usual optimal procedures for a real load data from the Southest Brazilian Company. A residual analysis of these forecasting models is given. In addition, the classical ARIMA and neural network models are also developed for the subject data, and the results are compared with the proposed models that we have introduced. In all cases the new models give better short-term forecasting results than the classical models.