These principles include 24 :.
Actors and their actions should be viewed as autonomous and independent units rather than as interdependent units. The links between actors should be viewed as channels for transfer of material and non-material resources. Social network models focusing on individuals view the structural environment as a network imposing certain constraints on individual actions. Social network models conceptualize structure social, economic, political, and so on as long-lasting patterns of relations among actors. Milgram concludes by describing how people of a population are connected.
Although the study was not subject to an evaluation process, his concept of the small world is widely adopted in social networks research to provide an explanation about how information spreads in the real world.
The small-world network has become one of the most widely used social network models. More recently, the influence of social networks has been found to be a key contributing factor to customer retention. Phadke and colleagues 21 developed a model that integrates SNA with traditional churn modeling concepts. The model was applied to a dataset of over half a million subscribers, provided by a large mobile network provider.
- نوشتن نقد و نظر;
- Shop with confidence.
- Surgery Revison!
The dataset contained customer call detail records. To compute social tie strength, the authors used three attributes: 1 The number of calls placed between two users; 2 the total duration of calls between two users; and 3 neighborhood overlap of the two connected users. The study found that users who make phone calls to numbers on a different network are also likely to churn in future in order to save costs.
Similarly, Verbeke and colleagues 22 conducted a study investigating the impact of social networks on customer retention. However, the latter differs from the former in that it uses both networked and non-networked customer-related information about millions of users. Although the studies mentioned above have contributed to knowledge of customer retention, they do not capture possible factors as to why customers make their decision to churn. ABMS is able to explore the dynamics of customer behavior. Over the years, researchers and industry practitioners have attempted to apply different techniques to understand customer behavior in the market place.
The ABMS approach is a typical example of one such technique. For example, carrying out an ABMS exercise with customers who enter a retail space. ABMS can be used in this context to derive insights into customer behavior within the retail store. ABMS is composed of two main activities: modeling and simulation.
Modeling is the process of representing real-life events into a model, while simulation is the process of executing the represented models such that they imitate the proposed system. ABMs are composed of agents and a structure for agent-based interaction. Agents can represent anything from a number of patients in a hospital to consumers of a product or service. ABMs are often characterized by rules and these rules define the behavior of agents in the system.
In such cases, a balance may be difficult to reach, making the ability to study the underlying system and the dynamics of the behavior imperative. ABM is distinct from traditional modeling approaches where characteristics are often aggregated and manipulated. Traditional modeling techniques are often suitable for their own purposes but they may not be able to provide adequate levels of detail when considering independent behaviors of agents. Although commonly used data-mining techniques for modeling consumer markets are powerful with regard to their purposes, they are generally not able to provide sufficient levels of detail with regards to the modeling of interdependent behaviors of consumers in a complex marketplace.
In addition, ABM is able to sufficiently represent interdependent systems even on a large scale i.
These promising approaches typically identify agents upfront, whereas here we uncover agents from the data analysis itself — in relation to determinants of interest. Domain complexity warrants a looser customer typing in order to adapt to new or changing data content and structure.
In the study of social behavior and interactions, ABM starts with a set of assumptions derived from the real world deductive , and produces simulation-based data that can be analyzed inductive. This paper applies the problem-solving design science paradigm as an overarching methodology in order to uncover techniques to rapidly create agent models from both the data analysis of the market and social interaction within it.
The problem being addressed is that markets are dynamic and models need to be speedily created in response. The next section provides a background on design science research DSR methodology. DSR is a multidisciplinary approach that primarily uses design as a research method or technique to solve a problem and learn from the process of solving that problem. Apart from its popular adoption in information systems, it is also widely used in disciplines such as education, engineering computer science, and healthcare.
Design science output or artifacts includes constructs, models, methods, and instantiations, while the natural science activities include build, evaluate, theorize, and justify.
Vijay K. Vaishnavi (Author of Design Science Research Methods and Patterns)
The four artifact types comprise:. Constructs: Articulate set of concepts that are used to describe problems within a domain and specify their solutions. Constructs also form the vocabulary of a discipline. Models: Define a set of statements which express relationships among constructs and represent real-world design activities in a domain.
Methods: A sequence of steps used to execute a task. These steps provide guidelines on how to solve problems with the use of constructs and models. Furthermore, methods can be described as a set of methodological tools that are created by design science and applied by natural science. Instantiations: The utilization of constructs, models, and methods to showcase an artifact in a domain.
They demonstrate the effectiveness of constructs, models, and methods. Instantiations provide working artifacts that can drive significant advancement improvement in both design and natural sciences.
A DSR methodology incorporates five stages of a design cycle to address design research problems. These phases are designed to aid sustainable development during the research and transfer knowledge from one iteration to the next iteration until a desired result is achieved. The next section explains the DSR processes. The DSR process follows a stepwise approach structured as five phases, shown in Figure 1.
Awareness of problem: The DSR process begins by identifying the problem under study. The identified problem may arise from multiple sources, such as the literature or current problems in the industry. The research problem needs to be clearly defined and articulated. The output of this phase is a formal or informal proposal for new research.
In this study, the core problem is identifying an effective approach to modeling changing customer behavior as a means to understand and improve customer retention. Included is the need to carry out experimentation in areas of little theory, such as the impact of social connections. A rich environment of internal and external data provide additional context, motivating a data-driven strategy.
Suggestion: Possible solutions about the research problem are explored and evaluated, leading to the acquisition of further insights to the domain under study. The specifications of the appropriate solutions to the research problem are defined.
Design Science Research Methods and Patterns: Innovating Information and Communication Technology
The output of this phase is a conditional design or representation of proposed solutions. This is articulated in diagrammatic and narrative form. Machine learning approaches are selected and tested using the dataset and R code. Development: This phase involves further developing and implementing DSR artifacts based on the suggestions from the previous phase.
During this phase, the CADET approach is further developed using a selected machine learning algorithm, refining R code to better generate the required outputs. Evaluation: Developed artifacts are analyzed and evaluated according to the criteria set awareness of problem phase. Deviations and expectations should be noted and explained. If the outcomes derived from the development or evaluation phase do not meet the objectives of the problem, the design cycle returns to the first phase, along with the knowledge gained from the process of the first round of work.
The phase may be iterated until the evaluation of the artifacts meets the solution requirements. The outputs of this phase should improve the efficiency and effectiveness of the artifacts.