International Journal of Information Technology & Decision Making

Uplift modeling combines machine learning and experimental strategies to estimate the differential effect of a treatment on individuals’ behavior. The paper considers uplift models in the scope of marketing campaign targeting. Literature on uplift modeling strategies is fragmented across academic disciplines and lacks an overarching empirical comparison. Using data from online retailers, we fill this gap and contribute to literature through consolidating prior work on uplift modeling and systematically comparing the predictive performance and utility of available uplift modeling strategies. Our empirical study includes three experiments in which we examine the interaction between an uplift modeling strategy and the underlying machine learning algorithm to implement the strategy, quantify model performance in terms of business value and demonstrate the advantages of uplift models over response models, which are widely used in marketing. The results facilitate making specific recommendations how to deploy uplift models in e-commerce applications.

This study proposes an advanced Modified KEmeny Median Indicator Rank Accordance (KEMIRA-M) approach based on stochastic evaluation process considering consistent weights to improve effective usage of KEMIRA-M. In the proposed approach, tasks related to the decision issue are performed by decision makers to ensure the understanding sufficiency of alternatives in terms of criteria more clearly. The weighting procedure of Analytic Hierarchy Process (AHP) is implemented in a stochastic manner benefited from discrete uniform distribution to provide obtaining consistent criteria weights considering median priority components. Therefore, different trials including different number of replications that shows the number of decision makers are performed and the most consistent weightings are determined for each trial in the stochastic process. In this way, the dependency to the limited numbers of decision makers and to determine criteria weights in a heuristic manner in KEMIRA-M is prevented. Additionally, the effect of the number of decision makers on criteria weightings and alternatives’ ranking process is shown. To obtain the most consistent weighting results, this stochastic process is utilized until acquiring approximate consistency ratios. The proposed stochastic KEMIRA-M approach is utilized to rank nine shopping malls (SMs) in Ankara in terms of technical criteria (TC) and universal design criteria (UDC). It was seen from the ranking results that the first SM (SM₁) is the best one.

The particle swarm optimization (PSO) algorithm is simple to implement and converges quickly, but it easily falls into a local optimum; on the one hand, it lacks the ability to balance global exploration and local exploitation of the population, and on the other hand, the population lacks diversity. To solve these problems, this paper proposes an improved adaptive inertia weight particle swarm optimization (AIWPSO) algorithm. The AIWPSO algorithm includes two strategies: (1) An inertia weight adjustment method based on the optimal fitness value of individual particles is proposed, so that different particles have different inertia weights. This method increases the diversity of inertia weights and is conducive to balancing the capabilities of global exploration and local exploitation. (2) A mutation threshold is used to determine which particles need to be mutated. This method compensates for the inaccuracy of random mutation, effectively increasing the diversity of the population. To evaluate the performance of the proposed AIWPSO algorithm, benchmark functions are used for testing. The results show that AIWPSO achieves satisfactory results compared with those of other PSO algorithms. This outcome shows that the AIWPSO algorithm is conducive to balancing the abilities of the global exploration and local exploitation of the population, while increasing the diversity of the population, thereby significantly improving the optimization ability of the PSO algorithm.

This study analyzes the affecting variables on the relation of sustainability practice-performance in the service supply chain. The identified factors (supply chain features) are classified in a hierarchical structure of Fuzzy Analytic Hierarchy Process (FAHP): Macro Environment Features, Service Provider Features, Supplier Features, Employee Features, and Customer Features. Then, the FAHP is applied to identify the importance of each factor in the hospital supply chain, as an example of the service sector. To extract and predict the factors’ interactions, the knowledge base of the fuzzy if-then rules is constructed. The developed Fuzzy Inference System (FIS) gives insights on the supply chain features with significant impact on the effectiveness of sustainability initiatives. So, by predicting and strategically managing the affecting factors, service supply chain managers can have more effective and sustainable performance. The fuzzy surfaces indicate that the features of the macro environment, especially the government regulations, are crucial factors influencing the movement toward sustainability.

Hotel competency is inherently intangible and multivariate. It involves a multiple-criteria decision-making problem. Particularly in the currently rapidly shrinking hotel market in Taiwan, what determines domestic chain hotel groups’ (DCHGs’) competence and survival involves more complex and multiple factors. A practical and effective tool is urgently required for making appropriate decisions. This paper thus proposes a combined consistent fuzzy preference relations (CFPR) and the VIKOR model, aiming to prioritize criteria and solution alternatives for hotel managers. In contrast to prior studies that have used mathematical programming, the model here is also tested using real-world hotel management cases and expert consultation. First, based upon the resource-based view, we propose 21 criteria and six dimensions as the determinants of DCHG competencies. Then, VIKOR is applied to produce the most appropriate alternatives with the corresponding weights obtained using the CFPR method. The combined method successfully manages the problems of linguistic ambiguity and consistency, determines the relative weights of the different factors and provides a ranking priority. The result is compared with some similar methods and is shown to be more useful and reliable. Finally, the verified model can be used to produce strategies. A decision-maker can make selection(s) from the solution formula. Our study may thus contribute to the hotel industry with the efficient decision-making tool of resource-based view (RSV) and two-phase methodologies.

In financial markets, there are situations where investors have the future stock prices according to the experts’ evaluations rather than historical data. Thus, the estimations of the stock prices contain much subjective imprecision instead of randomness. This paper discusses a portfolio investment with options in such a kind of situation. Treating the stock index price as an uncertain variable, we build an uncertain mean-chance portfolio model based on uncertainty theory and provide the equivalent form of the model. Furthermore, we make a comparison of the optimal expected return between portfolio investment with options and without options. An important conclusion is reached: The portfolio investment with options produces a no less expected return than that without options. In addition, we make sensitivity analysis and get two vital corresponding results. As an illustration, a numerical example is presented as well. The numerical results reveal that the options should be considered in portfolio investment. And the call option with maximum exercise price is most valuable per premium cost with the same exercise date.

In the era of big data, how to obtain useful knowledge from online news and utilize it as an important basis to make investment decision has become the hotspot of industrial and academic research. At present, there have been research and practice on explicit knowledge acquisition from news, but tacit knowledge acquisition is still under exploration. Based on the general mechanism of domain knowledge, knowledge reasoning, and knowledge discovery, this paper constructs a framework for discovering tacit knowledge from news and applying the knowledge to stock forecasting. The concrete work is as follows: First, according to the characteristics of financial field and the conceptual cube, the conceptual structure of industry–company–product is constructed, and the framework of domain ontology is put forward. Second, with the construction of financial field ontology, the financial news knowledge management framework is proposed. Besides, with the application of attributes in ontology and domain rules extracted from news text, the knowledge reasoning mechanism of financial news is constructed to achieve financial news knowledge discovery. Finally, news knowledge that reflects important information about stock changes is integrated into the traditional stock price forecasting model and the newly proposed model performs well in the empirical analysis of polyester industry.

The cloud model (CM) is an important tool to describe qualitative concept by the quantitative method, and the Maclaurin symmetric mean (MSM) can capture the interrelationship among the multi-inputs and it can generalize most of existing operators. In this paper, we firstly convert the uncertain linguistic variables (ULVs), which are easily used to express the qualitative information, to CM. Then, we combine the MSM with the CM, and propose the cloud MSM (CMSM) operator and cloud weighted MSM (CWMSM) operator. In addition, we explore some of their desirable features and develop a new approach to deal with some multi-attribute group decision-making (MAGDM) problems under the uncertain environment based on the proposed operators. Finally, by comparing with other approaches, an illustrative example is arranged to demonstrate the usability of the proposed method.

This study proposes an alternative method for evaluating the overall performance of hospitals in Taiwan. Specifically, three types of hospital activities, medical, pharmaceutical, and rehabilitation, are classified and connected using the parallel method. To bridge the difference between the performance evaluation methods adopted in hospitals and the evaluation method adopted by traditional enterprises, each activity is divided into two serial-type connected processes, which are the production and service processes. This study analyzes the performance of the six processes in three types of activities of three levels of hospitals, namely medical centers, regional hospitals, and district hospitals, by using multi-activity network data envelopment analysis (MNDEA) in Taiwan in 2012. The average overall efficiency was 0.41, while the average estimated efficiency for medical, pharmaceutical, and rehabilitation activities was 0.62, 0.34, and 0.27, respectively. At the hospital level, the average efficiency for medical centers, regional hospitals, and district hospitals was 0.59, 0.49, and 0.35, respectively. The production processes of medical and rehabilitation activities in Taiwan have higher average efficiency than their service processes, while the production processes of pharmaceutical activity have lower average efficiency than its service processes. These results provide a reference for upgrading the operation and management strategies for all hospitals in the country.

Usually, real-world problems involve the optimization of multiple, possibly conflicting, objectives. These problems may be addressed by Multi-objective Reinforcement learning (MORL) techniques. MORL is a generalization of standard Reinforcement Learning (RL) where the single reward signal is extended to multiple signals, in particular, one for each objective. MORL is the process of learning policies that optimize multiple objectives simultaneously. In these problems, the use of directional/gradient information can be useful to guide the exploration to better and better behaviors. However, traditional policy-gradient approaches have two main drawbacks: they require the use of a batch of episodes to properly estimate the gradient information (reducing in this way the learning speed), and they use stochastic policies which could have a disastrous impact on the safety of the learning system. In this paper, we present a novel population-based MORL algorithm for problems in which the underlying objectives are reasonably smooth. It presents two main characteristics: fast computation of the gradient information for each objective through the use of neighboring solutions, and the use of this information to carry out a geometric partition of the search space and thus direct the exploration to promising areas. Finally, the algorithm is evaluated and compared to policy gradient MORL algorithms on different multi-objective problems: the water reservoir and the biped walking problem (the latter both on simulation and on a real robot).