Historical Demand Data Research Articles

Data-driven vehicle rental and routing optimization: An application in online retailing

Due to limited self-owned vehicles, online retailers often struggle to meet high demands for deliveries, especially during large promotions. This study employs machine learning to tackle this challenge by shipping products and renting vehicles in advance. We explore a large amount of historical demand data, enabling accurate forecasting of demand information. It is then combined with an improved meta-heuristic algorithm named the Improved Discrete Whale Optimization Algorithm (IDWOA) to help online retailers make optimal decisions. The algorithm involves a discretization method and an effective perturbation strategy, along with information sharing, Cauchy mutation, and an elimination strategy. Experimental results demonstrate that our method can reduce costs by 14.78% compared to temporary vehicle rentals, and it significantly outperforms other comparative algorithms. Therefore, our study effectively integrates machine learning algorithms with an improved meta-heuristic approach, allowing for increased utilization of data-driven advantages to enhance the precision and efficiency of vehicle rental and routing optimization.

Enhancing Long-Term Water Quality Modeling by Addressing Base Demand, demand Patterns, and Temperature Uncertainty Using Unsupervised Machine Learning Techniques.

Water quality modelling in Water Distribution systems (WDS) is frequently affected by uncertainties in input variables such as base demand and decay constants. When utilizing simulation tools like EPANET, which necessitate exact numerical inputs, these uncertainties can result in inaccurate simulations. This study proposes a novel framework that leverages unsupervised machine learning, specifically a Gaussian Mixture Model (GMMs), to represent and integrate these uncertainties in the simulation. By classifying historical water demand into fuzzy clusters, the framework allows for certain linguistic inputs (e.g., "high" or "low" demand) to be used in water quality simulations. The framework also incorporates representative hourly demand patterns and temperature-dependent chlorine decay constants based on historical data correlations. Validations were conducted on the Anytown network using WNTR-EPANET, comparing simulated chlorine residuals with Validation data from 181 steady-state simulations. The simulation through the framework achieved a Jensen-Shannon Divergence (JSD) of less than 0.008 across all demand clusters, indicating high similarity between predicted and actual probability distributions . In comparison to other simulation scenarios tested, which exhibited increased variability (JSD > 0.18), the proposed framework demonstrated improved accuracy in representing chlorine residual distributions. The methodology is adaptable to other systems, if similar historical datasets containing key variables, such as flow rates and temperature, are provided. While the framework offers a more flexible and accurate approach to handling uncertainties in WDS, its effectiveness is contingent upon the availability of robust historical demand and temperature data for decay constant calibration.

Robust pricing and inventory decisions in ship‐from‐store omnichannel operations

AbstractThis work studies the deployment of the ship‐from‐store omnichannel strategy and the pricing and inventory decisions for an online retailer. Robust optimization models are constructed for the online‐only and the ship‐from‐store modes under a budgeted uncertainty set. The ARIMA model is used to predict the parameter values of the budgeted uncertainty set using historical demand data. The closed‐form optimal solution for the online‐only mode is obtained. The robust counterpart model for the ship‐from‐store mode is converted to a mixed integer quadratic programming model. Numerical studies are conducted to validate the theoretical results and to verify the effectiveness and practicality of the developed robust optimization solution approach. The results show that adopting a ship‐from‐store strategy may hurt the retailer's profit if a significant proportion of consumers are time‐sensitive with high travel cost. The ship‐from‐store strategy is optimal if it significantly boosts market growth.

One Method for Predicting Satellite Communication Terminal Service Demands Based on Artificial Intelligence Algorithms

This paper presents a traffic demand prediction method based on deep learning algorithms, aiming to address the dynamic traffic demands in satellite communication and enhance resource management efficiency. Integrating Seq2Seq and LSTM networks, the method improves prediction accuracy and applicability, especially for mobile terminals such as aviation and maritime ones. Unlike traditional approaches, it does not require extensive statistical data and can be generalized to real-world systems, providing stable long-term traffic demand predictions. This study utilizes real-world flight data mapped to corresponding satellite beams, allowing the precise prediction of beam-specific traffic demands. The results show that aggregating historical demand data for beams with similar trends achieves an average predictive Mean Squared Error (MSE) of 0.0007 and a maximum MSE fluctuation of 0.014, significantly outperforming predictions based on average values in terms of stability and accuracy. This novel solution for resource management in satellite communication ensures efficient and accurate long-term traffic demand predictions.

Shared autonomous vehicle operational decisions with vehicle movement and user travel behaviour

Shared Autonomous Vehicle (SAV) has many impacts on the transport development, such as saving parking space. However, SAV meets a huge challenge in terms of vehicle supply and user demand imbalance. The traditional mathematical optimization method cannot be well used due to the computational burden. Hence, this paper proposes a Reinforcement Learning (RL) based SAV relocation approach. First, two types of RL agents, car-based and zone-based agents, are developed as agents for vehicles and stations, respectively. Then, the RL scheme is trained by using historical demand data to facilitate real-time carsharing relocation. Finally, to compare the proposed two types of RL methods, three scenarios are used: small-scale, middle-scale, and large-scale networks. Solutions indicate that the enhanced zone-based method achieves an additional 146% profit compared to traditional threshold-based relocation strategies. The user travel behaviour impacts are provided by analyzing parking demand and travel movements among residential, industrial and commercial zones.

Energy efficiency and greenhouse gas emission reduction potential of solar PV and biomass-based systems for a food processing plant

The industrial sector is one of the most energy-intensive sectors, with a share of nearly 38 % (156 EJ) of all energy used globally in 2020. Historical energy demand and consumption data from a poultry processing plant have been analyzed to lower greenhouse gas emissions (GHG) and reliance on traditional energy sources. The study includes energy, economic, and emissions analyses for (a) solar photovoltaic (PV) systems, (b) biomass fuels, and (c) coal-fired with wood. Fixed, single, and two-axis tracking solar PV systems that are company-owned and funded by banks are considered. The historical energy consumption data of the plant shows that 52.36 % of energy demand is met by electrical energy and 47.63 % by thermal energy. The yearly electricity demand is 9,938 MWh, costs $1,192,560, and emits 5117.6 tCO2. Thermal energy demand is based on an annual use of 898 tonnes of coal and 71.6 tonnes of liquefied petroleum gas (LPG), which costs $ 143,681 and emits 2977.7 tonnes of carbon dioxide (tCO2). Feasibility of 1.3 MW, fixed axis company-owned, solar PV plant shows an annual electricity production of 1876 MWh, GHG emission reduction of 965.9 tCO2 with a payback period of 3.4 years. Biomass system delivers yearly fuel cost savings of $25,724, reducing 2784 tCO2 with a 1.9-year payback period. Co-firing biomass offers a 3.58 % fuel cost saving and 557 tCO2 with a 0.93-year payback period. It concluded that the wood fuel feedstock biomass option is most suitable renewable energy resource for such large-scale poultry facilities.

Multi-mode monitoring and energy management for photovoltaic-storage systems

The integration of photovoltaic generation systems and variable demand can cause instability in the distribution network, due to power fluctuations and the increase in reactants, particularly in the industrial sector. In response, photovoltaic units have been equipped with local storage systems, which eventually absorb power fluctuations and improve installation performance. However, during this procedure other functionalities that energy storage could provide are neglected. Consequently, this study provides a multi-mode energy monitoring and management model that enables voltage regulation, frequency regulation and reactive power compensation through the optimal operation of energy storage systems. With this objective, a smoothing control algorithm is developed that interacts with parameters of the electrical grid at the common connection point and also allows the compensation of reactive power based on an industrial demand profile. This strategy uses the Long short-term memory neural network of historical demand data prior to energy consumption with a relatively low RMSE of 1.2e-09. The results are previously validated in a development environment using a real-time OPAL-RT simulator and tests in the electrical Microgrid laboratory at the University of Cuenca. This configuration allows establishing a demand forecasting model that improves the supervision, automation and analysis of daily energy production. A series of results are provided and analyzed that demonstrate that the new tool allows taking advantage of the provision of multimode functionalities, achieving optimal voltage regulation and improving power quality by reducing the total harmonic distortion THD (V) and THD (I) indices by 0.5. % and 2 % respectively.

Data-driven contract design for supply chain coordination with algorithm sharing and algorithm competition

Supply chain members can intelligently learn their decisions based on historical data by using Machine-Learning (ML) algorithms. To coordinate the supply chain, the data-driven contract design problems for three contracts—buyback, quantity flexibility, and combined quantity flexibility and rebate—were investigated for a supply chain with one manufacturer and multiple retailers under algorithm sharing and algorithm competition. The problems were formulated as bi-level optimization models by introducing nonlinear mapping from historical demand data to ordering decisions and using ML algorithms to learn the mapping parameters. The bi-level optimization models were transformed into semi-infinite programming models and solved using the (nested) cutting plane methods. Empirical studies using data from two databases showed that algorithm sharing or algorithm competition, the type of contract used, and learning algorithms were the three factors influencing the performance of supply chain coordination when using a data-driven contract design. Algorithm sharing was found to be more beneficial to the supply chain members than algorithm competition in promoting supply chain coordination. An effective incentive mechanism, such as an individualized buyback ratio and a rebate from the manufacturer to the retailers with a good forecast performance, can encourage the retailers to participate in algorithm sharing and improvement.

Vegetable Automated Pricing Model Based on Particle Swarm Algorithm

Fresh grocery supermarkets face challenges such as the short shelf life of vegetables, quality degradation over time, and difficulty in selling products the following day. Considering these issues, they restock early in the morning without complete knowledge of the specific items or purchase prices, based on historical sales and demand data. To address this issue, this paper first visualizes the data to demonstrate a positive correlation between sales volume and time. Next, to address the issue of predicting the total daily restocking quantity and pricing strategy from July 1 to July 7, 2023, the document utilizes data from the previous three years to establish an ARIMA time series forecast model to predict the costs and sales volume of six vegetables for the first week of July 2023. Finally, to devise the replenishment quantity and pricing strategy for individual items on July 1, 2023, the paper filters the data for all vegetable items from June 24th to 30th, 2023, and combines it with the particle swarm optimization algorithm to obtain the results. This study is pragmatic, providing a feasible solution to the problem of automatic pricing for fresh vegetables.

A Study on Replenishment Decision of Vegetable Commodities Based on Time Series Forecasting and Linear Programming Models

The automatic pricing and replenishment decisions for vegetable products are based on the fact that vegetable products have relatively short shelf lives, and there is a wide variety of vegetable types for sale. Vegetable purchases typically occur in the early morning. Therefore, businesses often rely on historical sales and demand data for each product to make pricing and restocking decisions, even when they do not have exact information about individual item and purchase prices. This article is based on the data provided by the CUMCM to conduct relevant research.

Comparative analysis of battery energy storage systems’ operation strategies for peak shaving in industries with or without installed photovoltaic capacity

Battery energy storage systems can address energy security and stability challenges during peak loads. This study examines the integration of such systems for peak shaving in industries, whether or not they have photovoltaic capacity. The battery-sizing problem has been analyzed extensively. However, most analyses assume a specific battery operation strategy and ignore the impact of battery-charging schemes on system behavior. In this paper, the authors compare three different operation strategies for charging batteries in an industrial peak-shaving application based on historical demand data from a large electricity consumer in El Salvador. The three strategies are fast charging, time-based charging, and low-power threshold charging. The study analyzes the possible integration of a photovoltaic system with two different sizes for a range of battery sizes (from 250 to 1,500 kWh capacity), examining optimal peak shaving levels, economic savings, and battery degradation. Results indicate that fast-charging reduces monthly billing but degrades batteries faster. The estimated lithium-ion battery lifespan is 10-10.5 years, with a maximum difference of five months. These differences could affect the life cycle economics due to the high costs of battery replacement. The economic savings achieved by the peak shaving operation of the storage system are not enough to compensate the battery investment in this study. However, other case studies with different load profiles or other locations with more expensive electricity tariffs could make the adoption of these systems economically viable.

Integrated conservative profitability evaluation for newsboy-type product in an OBM group

ABSTRACT Manufacturers in competitive markets can enhance profit opportunities by transforming their business model to own branding and manufacture (OBM). Note however that in situations involving limited storage and production capacity, OBM groups must determine the optimal preparation quantity and frequently evaluate profitability in all channels. This is particularly important when dealing with newsboy-type (i.e. single-period) products. The achievable capacity index (ACI) can effectively measure and estimate the profitability of a newsboy-type product. However, the ACI is applicable only to measure a single product in a single retail point. In the current study, we developed an integrated achievable capacity index (IACI) to enable OBM groups with owned channels to measure the overall profitability with multiple independent normal demands. To estimate the true IACI, we derived the statistical properties of the IACI estimator based on the historical demand data. The lower confidence bound on IACI (LCBIA) was also presented to prevent overestimation due to sampling error. The LCBIA values for various values of parameters were summarized as the generic tables for the manager’s convenience pertaining to decision-making in conservative profitability evaluation. Finally, the applicability of the proposed scheme was assessed by applying numerical and sensitivity analyses to a real-world OBM example.

Robust inventory routing problem under uncertain demand and risk-averse criterion

Inventory routing problem (IRP), which plays an important role in implementing vendor-managed-inventory strategy, is to determine the optimal timing and quantity of products to be delivered as well as the optimal vehicle delivery routes from the vendor to retailers. Retailers’ demand is usually uncertain and their distributions are ambiguous. We utilize limited historical demand data of retailers to construct a series of scenarios to characterize demand uncertainties. Moreover, we develop a framework to construct ambiguity sets and describe the ambiguities in demand distributions. To cater to the new feature of IRP as being exposed to downside risk and balance the mean and risk level of cost, we incorporate the worst-case mean-conditional value-at-risk (M-CVaR) criterion into the objective function and present a new distributionally robust IRP model. We transform the proposed model into a tractable formulation based on duality theory. A case study demonstrates the feasibility of the proposed method. The results indicate that our model can provide a robust delivery solution for immunizing against the influence of ambiguous demand distributions. In addition, an out-of-sample performance analysis shows that the delivery solution provided by our model can effectively reduce the product shortage rate.

New product family demand planning: Addressing SKU‐level spread bias

AbstractNew product supply chain planning is challenging, primarily due to the lack of historical demand data. Rarely, however, do the academic literature or companies differentiate the demand forecasting process for new products from existing ones, despite their increased reliance on judgmental estimates. This research focuses on how judgmental errors lead to an under‐estimation of the difference between the highest‐ and lowest‐demand stock‐keeping units (SKUs), and consequently negatively impact supply chain planning for new product family introductions. A generalized empirical model and accompanying discrete event simulation are developed and applied to data from a major consumer packaged goods (CPG) firm during the launch of a new cosmetics product family. This application allows us to identify a focal type of judgmental error (identified as the SKU‐level spread bias) inherent to new product forecasting and to provide a new theoretical understanding of how this type of bias harms supply chain performance. Via an empirically driven theory‐building approach that iterates between the simulation outcomes and existing literature, SKU‐level spread bias is demonstrated to harm demand forecasts and, thereby, supply chain plans. Our unique theory‐building approach advances theory by identifying planner SKU‐level spread bias as a new source of bias that firms should seek to mitigate when introducing new product families.

Supply Chain Contracts in the Small Data Regime

Problem definition: We study supply chain contract design under uncertainty. In this problem, the retailer has full information about the demand distribution, whereas the supplier only has partial information drawn from historical demand realizations and contract terms. The supplier wants to optimize the contract terms, but she only has limited data on the true demand distribution. Methodology/results: We show that the classical approach for contract design is fragile in the small data regime by identifying cases where it incurs a large loss. We then show how to combine the historical demand and retailer data to improve the supplier’s contract design. On top of this, we propose a robust contract design model where the uncertainty set requires little prior knowledge from the supplier. We show how to optimize the supplier’s worst-case profit based on this uncertainty set. In the single-product case, the worst-case profit can be found with bisection search. In the multiproduct case, the worst-case profit can be found with a cutting plane algorithm. Managerial implications: Our framework demonstrates the importance of combining the demand and retailer information into the supplier’s contract design problem. We also demonstrate the advantage of our robust model by comparing it against classical data-driven approaches. This comparison sheds light on the value of information from interactions between agents in a game-theoretic setting and suggests that such information should be utilized in data-driven decision making. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.0325 .

A Unified Spatio-Temporal Inference Network for Car-Sharing Serial Prediction.

Car-sharing systems require accurate demand prediction to ensure efficient resource allocation and scheduling decisions. However, developing precise predictive models for vehicle demand remains a challenging problem due to the complex spatio-temporal relationships. This paper introduces USTIN, the Unified Spatio-Temporal Inference Prediction Network, a novel neural network architecture for demand prediction. The model consists of three key components: a temporal feature unit, a spatial feature unit, and a spatio-temporal feature unit. The temporal unit utilizes historical demand data and comprises four layers, each corresponding to a different time scale (hourly, daily, weekly, and monthly). Meanwhile, the spatial unit incorporates contextual points of interest data to capture geographic demand factors around parking stations. Additionally, the spatio-temporal unit incorporates weather data to model the meteorological impacts across locations and time. We conducted extensive experiments on real-world car-sharing data. The proposed USTIN model demonstrated its ability to effectively learn intricate temporal, spatial, and spatiotemporal relationships, and outperformed existing state-of-the-art approaches. Moreover, we employed negative binomial regression with uncertainty to identify the most influential factors affecting car usage.

Integrated Profitability Evaluation for a Newsboy-Type Product in Own Brand Manufacturers

Effective inventory management depends on accurate estimates of product profitability to formulate ordering and manufacturing strategies. The achievable capacity index (ACI) is a simple yet efficient approach to measuring the profitability of newsboy-type products with normally distributed demand, wherein profitability is presented as the probability of achieving the target profit under the optimal ordering quantity. Unfortunately, the ACI is applicable only to retail stores with a single demand. In the current study, we addressed the issue of measuring the integrated profitability of newsboy-type products sold in multiple locations with independent demand levels, such as own-branding-and-manufacture (OBM) companies with multiple owned channels. We began by formulating profitability in accordance with multiple independent normal demands, and then developed an integrated ACI (IACI) to simplify expression. We also derived the statistical properties of the unbiased estimator to determine the true IACI in situations where demand patterns are unknown. Finally, we conducted hypothesis testing to determine whether the integrated profitability meets a stipulated minimum level. For convenience, we tabulated the critical values as a function of sample size, confidence level, the number of channels, and the stipulated minimum level. One can make decisions simply by estimating the IACI based on historical demand data from all channels and then looking up the critical value in the corresponding tables. Consequently, the proposed methods make it possible for OBM managers to address integrated profitability evaluation, which is effective in deciding the optimal timing to pull unprofitable items from the shelves by looking up generic tables. Furthermore, we also performed numerical and sensitivity analyses for a real-world case to illustrate the applicability and some managerial implications of the proposed scheme.

Prediksi Penjualan Obat Dan Alat Kesehatan Terlaris Menggunakan Algoritma K-Nearest Neighbor

This vital health phenomenon raises problems related to identifying medicinal products and medical equipment that are most frequently prescribed by specialist doctors, and are in demand by patients, as well as efficient stock management. The main challenge faced by hospitals is the difficulty in predicting which medicines and health devices are most in demand. This research analyzes and predicts the best-selling medicines and medical devices based on historical sales and demand data. By adopting a machine learning approach using the K-Nearest Neighbors (KNN) algorithm, research can help hospitals optimize services, especially the availability of stock of medicines and health equipment. The analysis results provide deep insight into patient preferences and demand trends by specialist doctors, enabling smarter stock management adjustments. It is hoped that this solution will reduce stock shortages and waste of storage resources, contributing to more efficient healthcare services. In conclusion, this research shows that the KNN algorithm can provide intelligent solutions to overcome complex challenges in managing valuable health resources.

Electricity demand error corrections with attention bi-directional neural networks

Reliable forecast of electricity demand is crucial to stability, supply, and management of electricity grids. Short-term hourly and sub-hourly demand forecasts are difficult due to the stochastic nature of electricity demand. To improve the electricity demand forecasts, artificial neural networks are combined with attention-based bi-directional neural network as well as Error Correction methods, and Bayesian hyperparameter optimizations. As part of the hybrid VMD-CABLSTM-ANN-EC model, the intrinsic mode functions, representing distinct predictive features, are separated from electricity demand data using variational mode decomposition algorithm. For predicting each intrinsic mode function and respective error time series, convolutional neural networks-bi-directional long short-term memory algorithms are applied. In a second stage, error time series are split, followed by a model to predict the error series’ intrinsic mode functions, and a final integration stage to predict the error. The results of the proposed VMD-CABLSTM-ANN-EC model are benchmarked against decomposition-based and standalone models at substations in Queensland, Australia. The proposed VMD-CABLSTM-ANN-EC model outperformed all benchmark models. By splitting the predictor and target variables, it will be shown that the variational model decomposition method improves error corrections by revealing key features of historical demand data. Using error correction method in short-term electricity demand modelling can provide greater confidence in the prediction of electricity demand. The models can be used by energy providers for market analysis and insight research, to manage power failure risk, and make financial decisions.

A systematic analysis of design choices in short-term taxi demand prediction models

Accurate prediction of short-term taxi demand allows taxi services to strategically position idle vehicles in areas with insufficient supply, thus reducing idle times for taxis and waiting times for passengers. Towards this end, state-of-the-art approaches use Machine Learning (ML) models based on historic demand data, such as Long Short-Term Memory (LSTM), Fully Connected Feed Forward Neural Network (FCNN), Convolutional Neural Network (CNN), Graph Convolutional Network (GCN), or a combination of these. The majority of approaches use a grid representation of the environment to discretize the pickup locations of trips and time bins to order the temporal dimension of the large multi-target regression problem. While these models show good accuracy, many \ extit{design choices}, such as the size of the grid cells or the number of previous time steps inserted into a model, are made without justification or without proper tuning. This paper systematically investigates these design decisions. Our hypothesis is that considering them jointly rather than in isolation can improve accuracy. We investigate the influence of (1) the selected grid cell size, (2) the number of input maps, (3) the usage of time shift to enlarge the training data, and (4) the potential of transfer learning on the accuracy of short-term taxi demand prediction. We selected different network types and evaluated them on two large real-world datasets. The evaluation shows that our model outperforms state-of-the-art short-term taxi demand prediction models.