Search

Demand forecasting, a crucial aspect of anticipating future customer needs, involves using historical data to predict trends. With the rise of artificial intelligence (AI), companies are increasingly turning to machine learning algorithms to enhance accuracy in forecasting compared to traditional methods. This article delves into the application of machine learning algorithms in demand forecasting, specifically within the realm of supply chain management, addressing both long-term (4–5 years) and short-term (3–4 months) scenarios. The primary focus is on improving prediction accuracy by employing feature selection algorithms and various machine learning and deep learning approaches. Utilizing diverse algorithms, such as time series, traditional machine learning, and advanced deep learning techniques, the study aims to forecast demand for different timeframes. Evaluation metrics like Mean Absolute Percentage Error (MAPE) and Mean Percentage Error (MPE) are employed to assess the effectiveness of each model. The primary goal is to pinpoint the most effective algorithm tailored to a specific dataset. This empowers companies to make well-informed decisions and enhance their supply chain operations by leveraging precise demand forecasts. The results of this study hold the potential to empower decision-makers and practitioners by enhancing their forecasting capabilities. By integrating both forecast periods a more comprehensive and robust supply chain strategy is ensured.

Wind speed forecasting is essential for various domains, such as renewable energy generation, aviation, agriculture, and disaster management. However, wind speed is a complex and stochastic phenomenon that exhibits multiple stochastic patterns. Therefore, forecasting methods that can account for the dynamics and uncertainty of wind speed are required. The literature on wind forecasting classifies the time-scale into different categories, such as short-term, medium-term, and long-term. The data used for these categories is usually daily or sub-daily, such as hourly. This implies that high frequency data poses more challenges for the modeling. Machine learning (ML) methods have shown great potential in forecasting wind speed. However, ML methods may not be able to capture the complex and stochastic patterns of time series data, especially when there are multiple seasonal cycles. Therefore, it is often beneficial to use a preprocessing method like time series decomposition before applying ML methods. This article proposes a novel hybrid model that integrates machine learning and time series decomposition for forecasting wind speed. A multivariate seasonal trend decomposition method based on loess (MSTL) is used to preprocess the wind speed data before applying various forecasting methods. MSTL is a generalization of the popular STL method that can split the time series data into a trend component, multiple seasonal components, and a residual component. The seasonal components are considered as deterministic and forecasted using a seasonal naive method that repeats the most recent cycle. For the other non-seasonal components, the neural network autoregression (NNAR) method is employed. The same approach is followed for two common statistical methods: autoregressive integrated moving average and exponential smoothing state space model (ETS). This enables the comparison of the performance of machine learning and statistical forecasting methods based on time series decomposition. The MSTL-NNAR model is also contrasted with the NNAR model without preprocessing to assess the effectiveness of MSTL. The proposed model is applied to two wind speed datasets with different frequencies: daily and hourly. The model is evaluated for both long-term and short-term forecasts by comparing it with other models using various accuracy and bias measures. The results show that the hybrid model, MSTL-NNAR, outperforms all other models in terms of accuracy and bias for both types of data and both time horizons. The results also show that MSTL preprocessing significantly improves the performance of NNAR by reducing its error and bias. The results demonstrate the effectiveness of the hybrid model for forecasting wind speed data that are complex, nonlinear and high-frequency. The model can provide reliable forecasts for different time horizons and different types of data. The model can be useful for various applications, such as wind power generation, weather prediction, and environmental monitoring. [ABSTRACT FROM AUTHOR]

Demand forecasting, a crucial aspect of anticipating future customer needs, involves using historical data to predict trends. With the rise of artificial intelligence (AI), companies are increasingly turning to machine learning algorithms to enhance accuracy in forecasting compared to traditional methods. This article delves into the application of machine learning algorithms in demand forecasting, specifically within the realm of supply chain management, addressing both long-term (4–5 years) and short-term (3–4 months) scenarios. The primary focus is on improving prediction accuracy by employing feature selection algorithms and various machine learning and deep learning approaches. Utilizing diverse algorithms, such as time series, traditional machine learning, and advanced deep learning techniques, the study aims to forecast demand for different timeframes. Evaluation metrics like Mean Absolute Percentage Error (MAPE) and Mean Percentage Error (MPE) are employed to assess the effectiveness of each model. The primary goal is to pinpoint the most effective algorithm tailored to a specific dataset. This empowers companies to make well-informed decisions and enhance their supply chain operations by leveraging precise demand forecasts. The results of this study hold the potential to empower decision-makers and practitioners by enhancing their forecasting capabilities. By integrating both forecast periods a more comprehensive and robust supply chain strategy is ensured. [ABSTRACT FROM AUTHOR]

This paper covers the estimation of the water levels of Beysehir Lake, located in middle of Turkey, using the artificial intelligence (AI) such as the neural networks (NN) and the fuzzy logic (FL). The study considers the detailed investigation of the effect of the long-term estimate duration on the lake water level estimation. The analysed estimate ranges were 1 day, 30 days, 60 days and 90 days. The lake parameters such as the shortwave radiation, the lake total inflow rate, the lake total outflow rate and the past lake water levels constituted the input layer of the AI configurations. This study clearly showed that the estimate performance of the AI methods decreases with the increasing estimate range. It is also seen that the best estimate performance criteria are obtained by different AI methods for different estimate ranges. It is seen that the Generalized Regression Neural Network (GRNN) showed relatively superior performance compared with the other two artificial neural networks, i.e. the Radial Basis Function (RBF) and the Feed Forward Back Propagation method (FFBP), and the Adaptive Neuro-Fuzzy Inference System (ANFIS) method, for the long estimation ranges such as 60 and 90 days. The second overall best performance was obtained by FFBP.

This paper covers the estimation of the water levels of Beysehir Lake, located in middle of Turkey, using the artificial intelligence (AI) such as the neural networks (NN) and the fuzzy logic (FL). The study considers the detailed investigation of the effect of the longterm estimate duration on the lake water level estimation. The analysed estimate ranges were 1 day, 30 days, 60 days and 90 days. The lake parameters such as the shortwave radiation, the lake total inflow rate, the lake total outflow rate and the past lake water levels constituted the input layer of the AI configurations. This study clearly showed that the estimate performance of the AI methods decreases with the increasing estimate range. It is also seen that the best estimate performance criteria are obtained by different AI methods for different estimate ranges. It is seen that the Generalized Regression Neural Network (GRNN) showed relatively superior performance compared with the other two artificial neural networks, i.e. the Radial Basis Function (RBF) and the Feed Forward Back Propagation method (FFBP), and the Adaptive Neuro-Fuzzy Inference System (ANFIS) method, for the long estimation ranges such as 60 and 90 days. The second overall best performance was obtained by FFBP. [ABSTRACT FROM AUTHOR]

Many companies operating on the natural gas market use natural gas storage to balance production and transport capacities with major variations in gas demand. This paper presents an approach to predicting users' gas flow nomination in underground gas storage by different users. A one-year prediction horizon is considered with weekly data resolution. Basic models show that whereas for the great majority of users we can predict nomination based only on weather data and technical parameters, for some users additional macro-economic data significantly improved prediction accuracy. Various modeling techniques such as linear regression, autoregressive exogenous model and Artificial Neural Network were used to develop prediction models. Results show that for most users an Artificial Neural Network provides optimal accuracy, indicating the non-linearity of the relationship between input and output variables. The models developed are intended to be used as support for facility operation decisions and gas storage product portfolio modifications. [ABSTRACT FROM AUTHOR]

Abstract Long term electricity consumption forecasting has been extensively investigated in recent years in different countries due to its economic and social importance. In this context, the long term electricity consumption projections of a country or region are highly relevant for decision-making of companies and organizations operating in any energy system. In this paper, it is proposed a methodology that combines the bottom-up approach with hierarchical linear models for long term electricity consumption forecasting of a particular industrial sector considering energy efficiency scenarios. In addition, the Bayesian inference is used for model parameter estimation and, enabling the inclusion of uncertainty in the forecasts produced by the model. The model was applied to the Brazilian pulp and paper industry and it was able to capture the trajectory of the real consumption observed during the 2008–2014 period. The model was also used to generate long term point and probability distribution forecasts for the period ranging from 2015 until 2050. Highlights • Bottom-up Bayesian extension for the long-term electricity consumption forecasting. • The proposed model combines the bottom-up approach with hierarchical linear models. • The proposed model considers energy efficiency scenarios. [ABSTRACT FROM AUTHOR]

Long term annual electricity consumption forecasting is very important for country's energy planning. These forecasts are influenced by several factors (political, technological, social, environmental and economic), and brings with itself a high uncertainty degree in its results and difficulties in the evaluation of such factors over them. A methodology that eases to take into account these factors aiming improve the results and help understanding the electricity consumption annual trajectory till the forecast horizon is, therefore, very much useful and desired. So, we propose a modelling structure using the bottom-up approach to cope with these matters and to evaluate the trajectory of long term annual electricity consumption of a sector of the Brazilian industry up to 2050 considering energy efficiency (EE) scenarios. It is important to emphasize that Brazil is a developing country, and to build a bottom-up approach was a challenge, mainly due to the fact that this model is data intensive. In particular, this modelling was applied in the pulp and paper sector. The main goal was to consider technological diffusion scenarios in EE measures, and show the energy savings achieved. The results point an energy savings in the order of 25% when an actual scenario is considered. [ABSTRACT FROM AUTHOR]

Long term water demand forecasting is needed for the efficient planning and management of water supply systems. A Monte Carlo simulation approach is adopted in this paper to quantify the uncertainties in long term water demand prediction due to the stochastic nature of predictor variables and their correlation structures. Three future climatic scenarios (A1B, A2 and B1) and four different levels of water restrictions are considered in the demand forecasting for single and multiple dwelling residential sectors in the Blue Mountains region, Australia. It is found that future water demand in 2040 would rise by 2 to 33 % (median rise by 11 %) and 72 to 94 % (median rise by 84 %) for the single and multiple dwelling residential sectors, respectively under different climatic and water restriction scenarios in comparison to water demand in 2010 (base year). The uncertainty band for single dwelling residential sector is found to be 0.3 to 0.4 GL/year, which represent 11 to 13 % variation around the median forecasted demand. It is found that the increase in future water demand is not notably affected by the projected climatic conditions but by the increase in the dwelling numbers in future i.e. the increase in total population. The modelling approach presented in this paper can provide realistic scenarios of forecasted water demands which would assist water authorities in devising appropriate management strategies to enhance the resilience of the water supply systems. The developed method can be adapted to other water supply systems in Australia and other countries. [ABSTRACT FROM AUTHOR]

Abstract: Groundwater systems are in general characterised by non-stationary and nonlinear features. Modelling of these systems and forecasting their future states requires identification and capture of these underlying features that seem to drive these processes. Recently, wavelets have been used extensively in the area of hydrologic and environmental time series forecasting owing to its ability to unravel these aforementioned component features. In this paper, dynamic wavelet based nonlinear model (Wavelet Volterra coupled model) is tested for its ability to yield reliable long term forecasts of groundwater levels at two sites in Canada. The model results are compared with the results from other recent techniques like wavelet neural network (WA-ANN), Wavelet linear regression (WLR), Artificial neural network and dynamic auto regressive (DAR) Models. The results of the study show the potential of wavelet Volterra coupled models in forecasting groundwater levels in addition to being more versatile and simpler to use when compared with other competing models. [Copyright &y& Elsevier]

Abstract: Energy production and distributing have critical importance for all countries especially developing countries. Studies about energy consumption, distributing and planning have much importance at the present day. In order to manage any power plant or take precautions about energy subject, many kinds of observations are used for short, mid and long term forecasting. Especially long term forecasting is in need to plan and carry on future energy demand and investment such as size of energy plant and location. Long term forecasting often includes power consumption data for past years, national incoming per year, rates of civilization, increasing population rates and moreover economical parameters. Long term forecasting data vary from one month to several years. Some of the forecasting models use mathematical formulas and statistical models such as correlation and regression models. In this study, artificial intelligence is used to forecast long term energy demand. Artificial intelligences are widely used for engineering problems to solve and obtain valid solutions. Adaptive neural fuzzy inference system is one of the most famous artificial intelligence methods and has been widely used in literature. In addition to numerical inputs, Adaptive neural fuzzy inference system has linguistics inputs such as good, bad and ugly. Adaptive neural fuzzy inference system is used to obtain long term forecasting results and the results are compared to mathematical methods to show validity and error levels. In order to show error levels, mean absolute error and mean absolute error percentage are used. Mean absolute error and mean absolute error percentages are very common and practical methods in literature. The obtained error results, from 2003 to 2025, mean absolute error and mean absolute percentage error are 1.504313 and 0.82439, respectively. Success of Adaptive neural fuzzy inference system for energy demand forecasting is 99.17%. [Copyright &y& Elsevier]

Abstract: The Kohonen self-organization map is usually considered as a classification or clustering tool, with only a few applications in time series prediction. In this paper, a particular time series forecasting method based on Kohonen maps is described. This method has been specifically designed for the prediction of long-term trends. The proof of the stability of the method for long-term forecasting is given, as well as illustrations of the utilization of the method both in the scalar and vectorial cases. [Copyright &y& Elsevier]

The model presented in this paper integrates two distinct components of the demand for durable goods: adoptions and replacements. The adoption of a new product is modeled as an innovation diffusion process, using price and population as exogenous variables. Adopters are expected to eventually replace their old units of the product, with a probability which depends on the age of the owned unit, and other random factors such as overload, style changes etc. It is shown that the integration of adoption and replacement demand components in our model yields quality sales forecasts, not only under conditions where detailed data on replacement sales is available, but also when the forecaster's access is limited to total sales data and educated guesses on certain elements of the replacement process. [ABSTRACT FROM AUTHOR]

• Forecasting water quality variables in prawn ponds is considered. • An approach to introduce mean reversion in state space models is proposed. • The approach constrains deviant forecasts in long-term multi-step-ahead forecasts. • One can select which state space components should be mean reverting. • The approach allows for modelling short and long-term dynamics. The contribution of this study is a novel approach to introduce mean reversion in multi-step-ahead forecasts of state-space models. This approach is demonstrated in a prawn pond water quality forecasting application. The mean reversion constrains forecasts by gradually drawing them to an average of previously observed dynamics. This corrects deviations in forecasts caused by irregularities such as chaotic, non-linear, and stochastic trends. The key features of the approach include (1) it enforces mean reversion, (2) it provides a means to model both short and long-term dynamics, (3) it is able to apply mean reversion to select structural state-space components, and (4) it is simple to implement. Our mean reversion approach is demonstrated on various state-space models and compared with several time-series models on a prawn pond water quality dataset. Results show that mean reversion reduces long-term forecast errors by over 60% to produce the most accurate models in the comparison. [ABSTRACT FROM AUTHOR]