In the middle of the 1970s, decision analysts created an influence diagram (ID) with an easy-to-grasp, straightforward semantic. It is quickly replacing the decision tree, which has been severely criticized for its exponential development in complexity as more and more variables are added to the model. 

ID's direct applicability in team decision analysis stems from its ability to model and solve explicitly for imperfect knowledge-sharing among team members. In game theory, ID extensions depict the game tree differently. A decision-making scenario can be represented concisely in graphical and numerical form using an ID. It is an extension of the Bayesian network that allows for the modelling and solving of issues involving probabilistic inference and decision-making.

Value chain

The value chain is a common metaphor for the sequence of tasks that must be completed to create value for a project. Data collection, cleansing and wrangling, system integration, KPI analysis, and presentation of results are all part of this process, designed to provide business value in line with the organization's objectives. Data scientists are vital in every industry, yet individual departments rarely share the same responsibilities in large corporations. 

Requirement analysis, design, implementation, testing, and evolution are the "value chain"'s five main components. While data science can benefit from several different process models, determining the ideal one is impossible. The CSIRO, the big data company Pivotal, and writers Miller and Mork are just a few examples of those who have presented models of this value chain to fit the needs of a data science project. Their proposed concepts are similar but tailored to meet specific company requirements. 

The Influence Diagram

A data scientist needs to be familiar with the primary drivers of business choices to formulate and analyze a problem statement. Stakeholders' business intuition often matches the model outputs only if they need more data modelling and statistics expertise. 

Influence diagrams are graphic representations of how known and unknown factors influence the outcomes of business decisions. You can get a bird's-eye view of the following using an influence diagram.

• The benefits one can reap from constructing a model
• The missing pieces of data that are essential to the project
• The price of obtaining this new data

Components

• The Known Variable - When a data scientist begins work on a project, there are some metrics that they already know. Examples include critical data and attributes, existing performance indicators, and business intuitions. 
• The Chance Variable - The chance variable is any metric that can't be predicted before work begins. It's crucial to remember that the significance of a random factor will eventually be established. The value of this variable is unknowable at the outset of a project because competitor data can be acquired from external sources.
• The Decision Variable - A determinant is a symbol for a decision variable. It shows how various known and unknown factors can affect business decisions and the types of decisions that can be affected. Insurance quotes, new product development, and distribution channels are just a few examples of strategic business decisions that could be affected by the data science project's known and unknown variables.
• The Objective Variable - The desired result of a data science project is an example of an objective variable. The objective metric is part of a larger framework whereby other decision factors play a role and should be in sync with corporate objectives.

Conclusion

An influence diagram is a simple visual representation of a decision problem. It illustrates the essential factors as nodes of diverse forms and colours, including decisions, uncertainties, and objectives. It depicts their interactions as arrows.

The math theory behind influence diagrams is meant to make understanding how different options and prices work together easier. People often think of the impact diagram as a way to solve problems. It shows us how to identify the essential parts of a problem, the goals, the known and unknown factors, and how different choices affect each other to get to the same place.

Image source: Unsplash