Advances in Business and Management Forecasting: Volume 13

Machine learning methods have recently gained attention in business applications. We will explore the suitability of machine learning methods, particularly support vector regression (SVR) and radial basis function (RBF) approximation, in forecasting company sales. We compare the one-step-ahead forecast accuracy of these machine learning methods with traditional statistical forecasting techniques such as moving average (MA), exponential smoothing, and linear and quadratic trend regression on quarterly sales data of 43 Fortune 500 companies. Moreover, we implement an additive seasonal adjustment procedure on the quarterly sales data of 28 of the Fortune 500 companies whose time series exhibited seasonality, referred to as the seasonal group. Furthermore, we prove a mathematical property of this seasonal adjustment procedure that is useful in interpreting the resulting time series model. Our results show that the Gaussian form of a moving RBF model, with or without seasonal adjustment, is a promising method for forecasting company sales. In particular, the moving RBF-Gaussian model with seasonal adjustment yields generally better mean absolute percentage error (MAPE) values than the other methods on the sales data of 28 companies in the seasonal group. In addition, it is competitive with single exponential smoothing and better than the other methods on the sales data of the other 15 companies in the non-seasonal group.

In an omnichannel environment, customers switch channels from product discovery to eventual purchase decision strategically. Hence, the biggest challenge for retailers nowadays is how to operate an effective omnichannel strategy. To improve inventory operational efficiency, this chapter investigates the influences of price setting and customers’ return probability on inventory forecasting. Subsequently, we explore how retailers participate in providing appropriate information delivery and product fulfillment. Specifically, a stylized newsvendor model, which incorporates customers’ showrooming behavior, is developed to address retailers’ inventory problem. Furthermore, we compare the benefits of buy-online-and-pick-up-in-store (BOPS) and showroom strategy which originates offline but is completed online. Three main findings are obtained as follows: (1)online and offline inventory order quantities augment with the ascending of pricing offline and online, respectively. Meanwhile, the inventory decisions increase when customers’ return probability declines; (2) the implementation of showroom helps retailers expand their pure online market coverage than BOPS, while it reduces the total inventory quantity if the proposition of unit online inventory cost accounting for product price exceeds physical store; and (3) showroom strategy is more profitable than BOPS option as long as unit online inventory cost is small enough. In addition, we find this boundary where showroom increases total profit expands with the attenuating of return probability.

The service industry is a major component of the economy. Raw material, components, assemblies, and finished products are shipped between suppliers, manufacturers, distributors, and retailers. Accordingly, timely receipt of shipped goods is crucial in maintaining the efficiency and effectiveness of such service processes. A service provider offers an incentive to the customer by specifying a competitive target time for delivery of goods. Further, if the delivery time is deviant from the target value, the provider offers to reimburse the customer for an amount that is proportional to the value of the goods and the degree of deviation from the target value. The service provider may set the price to be charged as a function of product value. This price is in addition to the operational costs of logistics that are not considered in the formulated model. For protection against deviation from target due dates, the service provider agrees to reimburse the customer. The reimbursement could be based on an asymmetric loss function influenced by the degree of deviation from the target due date as well as product value. The penalties could be different for early and late deliveries since the customer may experience different impact and consequences accordingly. The chapter develops a model to determine the amount (price) that the provider should add to the cost estimate of the delivery contract for protection against delivery deviations. Such a cost estimate will include the operational costs (fixed and variable) of the shipment, to which an amount is added to cover the expected payout to customers when the delivery time deviates from the target value. The optimal price should be such that the expected revenue will at least exceed the expected payout.

The Bureau of Economic Analysis provides data from 1969 to 2016 regarding state-level and county-level unemployment costs. These data are used to construct least-squares estimations including linear growth, the persistence of business cycles, and the unique anomaly of the Great Recession. Each of these models is constructed for North Carolina data, including the state as a whole and each individual county in the state. The state and county models are compared for differences and insights.

Bankruptcy prediction has attracted a great deal of research in the data mining/machine learning community, due to its significance in the world of accounting, finance, and investment. This chapter examines the influence of different dimension reduction techniques on decision tree model applied to the bankruptcy prediction problem. The studied techniques are principal component analysis (PCA), sliced inversed regression (SIR), sliced average variance estimation (SAVE), and factor analysis (FA). To focus on the impact of the dimension reduction techniques, we chose only to use decision tree as our predictive model and “undersampling” as the solution to the issue of data imbalance. Our computation shows that the choice of dimension reduction technique greatly affects the performances of predictive models and that one could use dimension reduction techniques to improve the predictive power of the decision tree model. Also, in this study, we propose a method to estimate the true dimension of the data.

In recent years, the problem of classification with imbalanced data has been growing in popularity in the data-mining and machine-learning communities due to the emergence of an abundance of imbalanced data in many fields. In this chapter, we compare the performance of six classification methods on an imbalanced dataset under the influence of four resampling techniques. These classification methods are the random forest, the support vector machine, logistic regression, k-nearest neighbor (KNN), the decision tree, and AdaBoost. Our study has shown that all of the classification methods have difficulty when working with the imbalanced data, with the KNN performing the worst, detecting only 27.4% of the minority class. However, with the help of resampling techniques, all of the classification methods experience improvement on overall performances. In particular, the Random Forest, in combination with the random over-sampling technique, performs the best, achieving 82.8% balanced accuracy (the average of the true-positive rate and true-negative rate).

We then propose a new procedure to resample the data. Our method is based on the idea of eliminating “easy” majority observations before under-sampling them. It has further improved the balanced accuracy of the Random Forest to 83.7%, making it the best approach for the imbalanced data.

This chapter develops a regression model structure based on the 34 peer companies of Verizon and their associated financial performance variables. Based on the regression model developed, the compensation level of the CEO of Verizon is determined.

This chapter concerns itself with the development of a regression model for determining the executive compensation of the AT&T CEO. The data observations for this model consist of a list of 21 comparable companies selected by the compensation committee of AT&T, its institutional investors, and AT&T advisors. A set of 24 financial variables for each of the companies is compiled as the data source for the regression model.

In this chapter, we consider the model of call center incoming call forecasting and staffing-level optimization. We first present the structure of the model and how an agent-based modeling technique could enrich the decision rule and the model. A matrix layout is introduced to present the model so that it can be understood in an efficient way from the perspective of a programmer. The agent-based queuing model will be used in forecasting. We then utilize the bisection method and stepwise method to optimize the staff level to satisfy a target range service-level criteria. Call center management could use the model in practice for their management forecasting and optimization decision-making process in terms of how many agents they need to achieve the target business efficiency goal.