Decision support systems and sensitivity analysis for business decisions

Technology for Decision Making

Information is of key importance in any company. For any organization to succeed, high quality information has to be produced. Decision Support Systems (DSS) are information frameworks that are based on computers. The manner in which they are designed helps administrators to choose one solution to a problem from a variety of solutions. A DSS is an interactive information system that is based on computers that has people, models, procedures, databases, software, devices and telecommunications. These help those involved in the decision making process to solve business problems that are semi-structured or unstructured. Sensitivity analysis is useful for analyzing information. This helps us to settle on the right decision. In this paper, two methods of decision making will be deeply discussed. These are the Sensitivity Analysis and the Decision Support System (DSS). Below is an approach of the research paper.

Thesis Statement

The methodology to use in order to assist in decision making in a business can be technology-based or one that follows a strong analytical approach. The research paper discusses the better system, be it the Decision Support System (DSS) or the Sensitivity Analysis.

The paper is structured such that both procedures are explained autonomously in detail; followed by a comparative analysis of the circumstances under which one would work better. Several uses of the strategies in business will also be summarized. An opinion on the way in which these strategies are of help in an organization has also been provided, followed by the conclusion.

Decision Support System (DSS)

A Decision Support System (DSS) is a flexible, adaptable and interactive information system that uses computer technology. It uses the guidelines of decisions, models and model base together with a database that is comprehensive and specific bits of knowledge from the decision makers. This prompts decisions that can be implemented to solve problems that would otherwise not be solved through science models. Subsequently, a DSS supports the making and effectiveness of complex decisions (TRIPATHI).

Characteristics of DSS

i. Handles a lot of data like searches in databases.

ii. Gathers data, including that which has been stored internally and externally in mainframe networks and computer systems and processes it.

iii. Gives report and presentation flexibility to suit the requirements of the decision maker.

iv. Includes graphical and textual orientation such as tables, charts, trend lines and more.

v. Uses software packages that are advanced to perform comparisons and analysis which are complex and sophisticated.

vi. Provides the decision maker with a lot of flexibility to solve problems that are both simple and complex through the support of optimization and methodologies that are heuristic and fulfilling.

vii. Performs "what if" objective-seeking analysis.

Components of DSS

Schematic View of DSS

DSS applications can be made of the following subsystems (TRIPATHI).

1. Data Management System: A database is the central component of the database management system . This contains important, situation relevant data. It is overseen by software known as the "database management system" (DBMS). An interconnection can be made between the database management system and the firm data warehouse, an archive for data that is important in making corporate decisions.

2. Model Management System: The model subsystem provides a wide range of models to decision makers and helps them in the process of making decisions. The model subsystem management software (MBMS) can be a part of the model subsystem. It coordinates model use in DSS. External data storage can also be linked to this component.

3. Knowledge-Centered Management System: This system can either perform as an independent component or support all additional systems. It increases the decision maker's intelligence through provision of its own. It can be interconnected with the organization knowledge base, the repository of information in the organization.

4. User Interface System: Also referred to as the management of dialog facility, the user interface permits users to get information through interaction with the DSS. Two capabilities are required by the user interface; the action dialect that.

Sensitivity Analysis (SA)

Sensitivity Analysis is the examination of how the variety in a model's output can be allocated to varying variation sources. Subsequently, the concepts of model and uncertainty are firmly connected to sensitivity analysis. Initially, SA was developed to handle uncertainties in the parameters of the model and input variables. The concepts have extended their boundaries to include model conceptual uncertainty. The study has included the expectations, ambiguity and specifications related to models over time. The Bayesian outline is used by arithmeticians to give a natural depiction of uncertainties on models and in variables and restrictions.

Sensitivity analysis is important in decision making. It can be used as an alternative of "What If?" analysis. Sensitivity analysis involves differentiating a model's input constraints within a stated area and evaluating the results. It also looks at methods that can be used to bring about a result from the various changes in the causes. Sensitivity Analysis works to find the level of change needed on the input data in order for the production recorded by linear programming to remain constant. From this, we can tell the level of sensitivity of the data given. If there is a big change in the prime solution for a model, caused by a little modification on the input; then the problem can be said to be less vigorous. However, if the little modification has not much effect on the prime solution, it can be said to be more vigorous. The latter has lower sensitivity to the modifications on input (L & Sawant, 2014).

Models are meant to mimic or estimate processes and systems which have fluctuating complexity and different forms; such as economic, social, physical or environmental. Those experiments that are deemed hard or unachievable can be mimicked through simulation modeling, which has been emphasized quite well in Rosen's ratification of modeling. (Rosen, 1991) talks about Aristotle's classification of connectedness. She states that the world is motivated by efficient and measureable cause whereas the model is motivated by formal connection. The two are connected through "decoding" (from model to world) and "encoding" (from world to model). The inside "model" and inside "world" causativeness is dominant, whereas decoding and encoding have no demand. They are simply used as objects of the art of the model designer. Decoding and encoding are the basis and principles of modeling. Models are designed with anticipation that the decoding process will create awareness for the world. This cannot be achieved unless the doubtfulness in the facts from the model (the object for decoding) is keenly assigned to the uncertainty from encoding (Center, 1999).

Applications for Sensitivity Analysis

Model designers can use SA to find out

i. the similarity of the model with the subject being studied,

ii. the value of the description of the model, iii. major aspects that lead to the output fluctuation

iv. the area in the space of aspects of input which lead to maximum model variation

v. optimum areas within the range of features used in a successive standardization study

vi. Relations between aspects.

Making Use of Decision Support System -- Business Case 1

A system was established for a car maker in the U.S., to which cars exceeding 1 million are returned from rentals or leases annually. The cars belong to the manufacture, and the issue is through which means these cars can be most successfully given out to hundreds of auction places all over the U.S. They are of different mileage, make, choices and damage, among other factors. These factors are among the determinants of the selling price of the cars in every auction. Our main problem was to give out the most possible number of cars to the auction sites. This would help maximize the sales profits. The procedure of coming up with prime approvals entails numerous contemplations, which range from estimation of price for different makes of cars in different places, to reduction in price and bulk effects, to issues of shipping. A million cars in a year equal around 4,000 cars in 1 working day. Therefore, a remarketing department has to make 4,000 resolutions, daily, on the auction site expected to maximize the selling price per car. Moreover, as a result of bulk effects, allocation of cars to the auction sites is greatly interconnected, making it impossible to make these cars consecutively (Michalewicz, Schmidt, Michalewicz, & Chiriac, 2005).

So, the problem can in one word be defined as volume. It is more economical to transport many cars at a go than it is to ship one or a few. Other than shipping expenses, other factors must be considered, such as shipping and auction schedule, the bulk effect, risk and insurance (due to theft or damage while transporting), and devaluation. The bulk effect comes in with increase in the number of alike cars on sale. If many similar cars are sent to one auction site (which may be appropriate if it has the most desirable price) the bulk will bring about less money for each car. Scheduling, too, is a big issue. There is a usual sales day for each auction, for instance as at 10 a.m. every third Tuesday. Hence, supposing there are 20 cars to be transported to one auction place, the shipping period is 10 days, and the following auction will be in 11 days. In case of any delay in delivery, the firm may not make it to the auction. It would have to leave the cars at the car lot in the auction place for about a fortnight.

To handle this problem, the company came up with an intelligent scheme which has numerous units. The units are optimization, adaptation and prediction elements. The prediction element has a number of parts. After setting of the default least price, the other elements alter the price and come up with the final expected price (Michalewicz, Schmidt, Michalewicz, & Chiriac, 2005).

i. Creation of the default base price. Through frequently updated data from the Black Book, this element identifies the price depending on make, region, year, mileage and model.

ii. Zip-code-related model/make tuning. Makes use of historic facts, such as records of sales, to regulate the price of a particular model or make.

iii. Zip-code-related group and color tuning. It stipulates alterations depending on the group or color of the car, be it compact, midsized or luxury.

iv. Mileage alteration. It depends on the car' make, mileage category, model and model-year period; for instance, if the 2006 model is introduced (e.g. in August 2005) the year age of the 2005 model is 1.

v. Seasonality alteration. It alters the price on the basis of the vehicle's model year, model, region and make. It does not consider the zip-code. The element computes the seasonality alteration through a daily devaluation rate, which is given per month.

It also computes the overall alteration by joining the daily alterations from the time the default least price is set to the time the sales are forecasted.

Making Use of Sensitivity Analysis -- Business Case 2

The managers of the project are in charge of implementing the project and also the project's deliverables. It is highly essential to evaluate the risk during the initial stage of the project. This will make it easy for risk to be managed, and for incidence planning to be done. Risk exposure is vital for identifying the risks at the initial stage. The process of analytical hierarchy was a procedure I found valuable for brainstorming and mostly for the project managers to devote their time. This would give an order that if these procedures would be used in the software business, they would minimize the loss brought about by inadequate outcome. Hence, it would not be necessary to use any risk management steps when implementing the projects, for delivery of 100% outcome to consumers. The managers of projects need to see past the practices of risk analysis towards the process of managing risk, since in the long run, it is the manager of the project who is the risk manager. In addition, it is recommended that the suitable practices of companies are adopted by the software business to advance the procedure, and for the managers of the project to devote their time measurably in these practices (Makhani, Khan, & Soomro, 2010).

Risk Prioritization:

In a scenario in the real world, the possibility of identifying numerous risks is common. It might therefore take a lot of time to investigate each one of them. This is why project managers realize the need for risk prioritization and risk analysis. Therefore, risk exposure is the most efficient and effective technique. However, estimating the probabilities and the loss connected to the unfulfilling outcome while managing the project is the problem it comes with. The whole process of risk analysis includes prototyping, benchmarking, and simulation. This provides more accurate estimations than the others. However, they are very costly and consume a lot of time.

Risk management infrastructure should include this analysis structure and it must fit into the matrix organization that has been discussed above. (Morris, 1988) states that generally, key projects start with a centralized structure, get decentralized at some point, and become centralized as they end. In the whole decentralized phase; there is need for a big management superstructure to maintain the integrity of the project. As much as there is no general structure for all the projects, (Busby, 1992) suggests very helpful pointers. Ireland and Shirley (Ireland & VD, 1986) also comment on a risk management system that is integrated.

Sensitivity Analysis and their Contribution to Business Success

Sensitivity analysis is widely used for the provision of a selection basis. It can assist in the designing of parameters through the observation of the characteristics and responses of the system in regard to predetermined valuables. This helps to identify the factors with the most impact as well as detection of bottlenecks in the system. There are two types of sensitivity analysis that can be used:

i. Nonparametric sensitivity analysis- it analyzes the responses of the system upon the addition or removal of a component or when the system model is modified.

ii. Parametric sensitivity analysis- it studies the behaviors of the system in regard to the varying input parameters.

The parametric sensitivity analysis is used to assess the performance of the system and its availability or reliability when the parameters in the different systems are changed (Nguyen, Min, & Park, 2015).

Examples of DSS helping businesses in Decision Making

Several organizations have incorporated decision support that is computerized into their daily operational activities such as monitoring performance. Managers often download and study sales data, generate feedback and then study and evaluate those reports. Through the use of DSS, managers can perform tasks like resource allocation, revenue projection and scenario evaluation. Data warehouses can come up with one version of the truth to be used in advanced analysis and reporting. Executive scorecards and dashboards have been adopted by more managers for operation tracking and supporting the making strategic decisions from their workstations (Power, 2016).

Better decision support is required due to changing decision-making environments, limitations on the decision makers and managerial requests. Creating a decision support system should be put into consideration when;

There is a possibility of improving decision quality through good information

The need for computerized support is recognized by the potential users of DSS

Making Managers Use the Decision Making Tools

You may wonder what is done by decision support systems to actually assist their users. How do they really impact on the users? The answers to such queries, according to my survey, proved elusive. This is attributed to the users valuing the systems for other reasons other than the intended ones. Apparently, there are ranging purposes existing for these systems. Majority of the decision support systems not only share the aim of standard EDP systems, but also address various managerial concerns that include facilitating problem solving, improving interpersonal communication, increasing organizational control and fostering individual learning (Alter, 1976).

Interpersonal communication can be affected by these systems in two ways: by providing tools for persuasion for individuals and also by providing a vocabulary and a discipline to an organization thus facilitating negotiations past subunit boundaries.

Aiding Communication

Managers require the assistance of the decision support systems to negotiate beyond organizational units. This is when they standardize the process mechanics and also provide a common conceptual ground to enable decision making.

Managers, during my survey, frequently commented on the importance of consistent definitions and formats to communication, more so among individuals in different organizational units, for instance different divisions or departments. The development of these formats and definitions in several of instances was a lengthy and occasionally arduous task. The accomplishment of this survey was a gradual process over a number of years. However, it was also considered to be one of the major contributions of the systems. For example, establishing beforehand the various outcomes of the decisions made by different people separately through filtering of their decisions via a single model was among the purposes of my sample's model-oriented systems. The system was an implicit arbiter between varying goals of different departments in these cases. People are able to use the model to indicate the effects of the proposal of one group to the other or on the total outcome instead arguments based on their own divergent viewpoints, production, marketing, and financial. The result will be issues being clarified and expedition of the negotiation process (Alter, 1976).

Adding Value to Users

According to my study, the implementers of several of the successful systems felt the importance of going through the motions to present a cost/benefit rationale hence attributing a dollar value to personal effectiveness. However, just like other people, they did not believe in these numbers. Management usually makes the decision to proceed based on the proposed system making sense and its capability to be of a beneficial impact on people's interactions and/or decision making processes. Obviously, monetary savings are worthwhile and also crucial rationale required in developing the computer systems. Nevertheless, it ought to be clear at this juncture that the EDP-style assumption that the systems ought to be justified in these terms is not sufficient in the field of decision support systems (Alter, 1976).

Caution Steps

There are great risks involved in developing a system based on the fact that someone thinks it is sensible to do so, especially if the particular individual is not using the systems directly. My first, second, and fifth example of systems citations indeed began in this way. These systems, before being repositioned as actual systems, its users may be required to be more effective if the firm faced a lot of resistance. The common tendency for technical people to pay more emphasis on the "technical beauty" of a system or idea is a general problem. Furthermore, they try to give assurance that nontechnical people for some reason will see the light and thus have the ability to figure out how to solve business problems using this system. In the history of nearly every unsuccessful system sampled there exists this sort of over optimism (Alter, 1976).