Combined Forecasting Approach Research Articles

Combined forecasting approach for product quality based on support vector regression and gray forecasting model

Forecasting product quality by incorporating customer satisfaction perception factors is an intriguing research area, which can promote the sustainable development of enterprises. To address small-sample random time series data, this study proposes a combined forecasting approach (CFA) for the product quality index that considers perception factors. The proposed approach is based on the support vector regression (SVR) and an improved gray forecasting model (GFM). First, the study constructs a system of perception factors related to defect parts per million (DPPM). Then, the key perception factors (KPF) are selected using the gray entropy relational degree, which is derived from gray relational analysis and information entropy. Then, a multivariable GFM is proposed based on the weighted Markov and the derived form of the gray model to reduce the forecasting error. Finally, a CFA is constructed considering KPF and optimized based on the SVR and the proposed GFM to forecast the DPPM. A case study of liquid crystal display is conducted to demonstrate the feasibility of the proposed CFA. The forecast error of the proposed CFA is 3.2%, which is better than those of GFM, SVR, and ARIMA (4.01%, 6.21%, and 9.89%, respectively). The comparison and discussion of methods demonstrate the superiority of the proposed approach for forecasting product quality.

Incoming Call Forecasting Framework for a Network Service Provider Call Center during Mobile Expo Events

An accurate forecasting model for call center operations allows a company to optimize the number of staffs needed. However, when a special event is planned, the call volume and pattern during that event can be expected to be different from those reflected in the historical database, and consequently, the accuracy of current forecasting methods may degenerate. Thus, a framework that can help to provide a highly accurate forecasting model in such situations is needed. In this paper, a conceptual framework for forecasting incoming calls during mobile expo events is proposed. The framework comprises four main steps: defining data types, verifying time series forecasting methods based on the defined data types, defining the forecasting model and forecasting the incoming calls. This approach can assist in systematically selecting an appropriate forecasting model when a mobile expo event is arranged. Experimental results show that this framework helps to select the appropriate forecasting model for each mobile expo event organized over one year. For the first and third mobile expo event periods, a mixed-period model with an individual forecasting approach, provides the best result, whereas in the second period, a combined forecasting approach for nonevent and mobile expo event periods, is the most appropriate.

Combined Forecasting Model of Cloud Computing Resource Load for Energy-Efficient IoT System

Cloud computing is generally considered as a special energy-efficient form for the Internet of Things (IoT) resource usage. Dedicated server systems for cloud services, better capacity utilization and economies of scale because of the use of larger and more energy-efficient data centers are the reasons why cloud solutions typically use less energy than traditional on-premise systems. To scientifically and rationally configure the hardware and software resources of the cloud computing, the research on forecasting a cloud computing resource load becomes a research focus. However, the widely-used single forecasting model cannot contain all the characteristics of the cloud computing resource load sequence, resulting in inaccurate forecasting results. In this paper, a combined forecasting approach of cloud computing resource load based on wavelet decomposition is proposed, which combined the grey model and cubic exponential smoothing model. It can well preserve details and reduce noise. Firstly, the cloud computing resource load sequence is decomposed into several frequencies by the wavelet decomposition method. The decomposed load sequences with different characteristics are divided into different resolution scale subspaces in deferent frequencies. The noise of the load sequences is reduced by the wavelet threshold denoising method. And then, the load sequences are reconstructed according to the wavelet coefficients. The reconstructed load sequence not only contains less noise but also reserves detailed information. Consequently, it is closer to the real data and more regular. Experimental results show that our proposed combined forecasting model with wavelet decomposition can provide more accurate forecasting results than each single forecasting model or the combined forecasting model without using the wavelet decomposition method. Thus, our proposal is demonstrated to be efficient for forecasting the cloud computing resource load and helping to reduce energy consumption.

Prediction of a service demand using combined forecasting approach

Forecasting facilitates cutting down operational and management costs while ensuring service level for a logistics service provider. Our case study here is to investigate how to forecast short-term logistic demand for a LTL carrier. Combined approach depends on several forecasting methods simultaneously, instead of a single method. It can offset the weakness of a forecasting method with the strength of another, which could improve the precision performance of prediction. Main issues of combined forecast modeling are how to select methods for combination, and how to find out weight coefficients among methods. The principles of method selection include that each method should apply to the problem of forecasting itself, also methods should differ in categorical feature as much as possible. Based on these principles, exponential smoothing, ARIMA and Neural Network are chosen to form the combined approach. Besides, least square technique is employed to settle the optimal weight coefficients among forecasting methods. Simulation results show the advantage of combined approach over the three single methods. The work done in the paper helps manager to select prediction method in practice.

A combined forecasting approach with model self-adjustment for renewable generations and energy loads in smart community

The short-term forecasting of wind power, photovoltaic (PV) generation and loads is important for the secure and economical dispatching of smart community with smart grid. Considering the smart community has plenty of distributed generations, here, a concept of net load is defined as the active power difference between renewable generations (wind and PV power) and loads. Then, a combined forecasting approach, which enables to build a real-time forecasting model with parameters self-adjustment, is proposed for the forecasting of the net load in smart community. Compared with the traditional forecasting methods such as support vector machine (SVM), the proposed approach can wavily optimize the parameters of the forecasting model. Besides, an optimal method named Grid-GA searching is developed to reduce the computation time during the forecasting. Therefore, it can improve the forecasting accuracy even if there is a great of uncertainty component in wind power, PV generation and loads. Detailed case studies give a contrastive analysis of the traditional and the proposed forecasting approach. The results show that the proposed approach has the capability of self-adaption on the fluctuations of wind and PV power, and can effectively improve the forecasting accuracy and efficiency.

A novel combined forecasting model for short-term wind power based on ensemble empirical mode decomposition and optimal virtual prediction

As one of the most promising renewable energy, wind energy plays a vital role in optimizing the configuration of energy resources in power system nowadays. However, wind generation with the intermittent and uncertain characteristics has brought new challenges for the integration of large-scale wind power into power system. Consequently, the accurate forecasting of wind power is the most effective and applicable solution to meet the challenges. A novel combined forecasting approach is proposed by integrating the ensemble empirical mode decomposition (EEMD) technique and the combination of individual forecasting methods based on optimal virtual prediction for the purpose of improving the short-term wind power prediction performance. There are three steps in this presented approach. First, EEMD is adopted to decompose the original wind power series into a number of intrinsic mode functions (IMFs) and a residue. Second, the prediction of each IMF is achieved by using four individual methods, and the prediction of the residue is obtained from the nonlinear grey Bernoulli model based on particle swarm optimization. Finally, the combined forecasting model based on optimal virtual prediction is developed, and the weight matrix in this model is optimized by a self-adaptive differential evolution algorithm, which aims to minimize the forecasting errors at the virtual prediction points. The real wind power data from a wind farm in China are used to verify the performance of the proposed model, and the simulation results show that this model has demonstrated the optimal forecasting accuracy and robustness compared with other forecasting models, which is a promising alternative for short-term wind power forecasting.

Improving short-term demand forecasting for short-lifecycle consumer products with data mining techniques

AbstractToday’s economy is characterized by increased competition, faster product development and increased product differentiation. As a consequence product lifecycles become shorter and demand patterns become more volatile which especially affects the retail industry. This new situation imposes stronger requirements on demand forecasting methods. Due to shorter product lifecycles historical sales information, which is the most important source of information used for demand forecasts, becomes available only for short periods in time or is even unavailable when new or modified products are introduced. Furthermore the general trend of individualization leads to higher product differentiation and specialization, which in itself leads to increased unpredictability and variance in demand. At the same time companies want to increase accuracy and reliability of demand forecasting systems in order to utilize the full demand potential and avoid oversupply. This new situation calls for forecasting methods that can handle large variance and complex relationships of demand factors.This research investigates the potential of data mining techniques as well as alternative approaches to improve the short-term forecasting method for short-lifecycle products with high uncertainty in demand. We found that data mining techniques cannot unveil their full potential to improve short-term forecasting in this case due to the high demand uncertainty and the high variance of demand patterns. In fact we found that the higher the variance in demand patterns the less complex a demand forecasting method can be.Forecasting can often be improved by data preparation. The right preparation method can unveil important information hidden in the available data and decrease the perceived variance and uncertainty. In this case data preparation did not lead to a decrease in the perceived uncertainty to such an extent that a complex forecasting method could be used. Rather than using a data mining approach we found that using an alternative combined forecasting approach, incorporating judgmental adjustments of statistical forecasts, led to significantly improved short-term forecasting accuracy. The findings are validated on real world data in an extensive case study at a large retail company in Western Europe.

Data-driven reduction targets for a highway safety plan

A safety plan or policy should have traffic safety targets. Setting the targets is challenging but is possible through data-driven approach. This paper describes a study that forecast fatalities and injuries under three scenarios, examined the probabilities of meeting different reduction targets, and proposed realistic targets for Virginia’s safety plan. The study combined statistical forecasting approach, inputs from safety experts, and reported effects of safety improvement measures to come up with ambitious yet achievable targets. Realistic targets for the safety plan in Virginia by 2010 are a 10% reduction in fatalities and a 5% reduction in injuries. These targets assume enactment of a primary safety belt law and deployment of crash countermeasures comparable to 20–30% engineering treatments.

An overview of accident forecasting methodologies

Accident forecasting is designed to help decision making and planning before casualty and loss occur. The main purpose of this paper is to gain insight into the general principles and methodologies for accident forecasting as well as current trends and future perspectives. In this paper, seven major methods are identified and discussed, including scenario analysis, regression method, time-series method, Markov chain method, grey model, neural networks and Bayesian networks. Based on a series of comparative analyses on their capabilities and limitations, some conclusions regarding the model selection and improvement are drawn. This paper argues that an individual method cannot build realistic forecasting model, and future perspectives should focus on the combined forecasting approach. The combination of non-linear model such as grey model and neural networks shows good performance in non-linear problems and has great potential in accident forecasting. An example is presented to illustrate the usefulness of combined forecasting.

A combined forecasting approach based on fuzzy soft sets

Forecasting the export and import volume in international trade is the prerequisite of a government’s policy-making and guidance for a healthier international trade development. However, an individual forecast may not always perform satisfactorily, while combination of forecasts may result in a better forecast than component forecasts. We believe the component forecasts employed in combined forecasts are a description of the actual time series, which is fuzzy. This paper attempts to use forecasting accuracy as the criterion of fuzzy membership function, and proposes a combined forecasting approach based on fuzzy soft sets. This paper also examines the method with data of international trade from 1993 to 2006 in the Chongqing Municipality of China and compares it with a combined forecasting approach based on rough sets and each individual forecast. The experimental results show that the combined approach provided in this paper improves the forecasting performance of each individual forecast and is free from a rough sets approach’s restrictions as well. It is a promising forecasting approach and a new application of soft sets theory.

Forecasting in Short-Term Planning and Management for a Casino Buffet Restaurant

SUMMARY Demand fluctuation accounts for an important consideration in a restaurant's daily operational decisions. Good short-term planning and management require accurate forecasts of daily demand. The objective of this study is three-fold: (1) to apply, evaluate, and compare different methods of forecasting customer counts for an on-premises buffet restaurant of a local casino in Las Vegas, (2) to describe and propose a combined forecasting approach for this casino buffet restaurant, and (3) to explore the concept of revenue and capacity management for this buffet restaurant. Eight forecasting models were tested and evaluated by two common error measures. The results suggest that a double moving average model was the most accurate model with the smallest MAPE and RMSPE. Extensive discussions on forecasting and planning/management in buffet operations are provided along with recommended future research.