Making Sense of Data: A Practical Guide to Exploratory Data Analysis and Data Mining

2.3 DELIVERABLES

It is also important to identify the deliverables of the project. Will the solution be a report, a computer program to be used for making predictions, a new workflow or a set of business rules? Defining all deliverables will provide the correct expectations for all those working on the project as well as any project stakeholders, such as the management who is sponsoring the project.

When developing predictive models, it is useful to understand any required level of accuracy. This will help prioritize the types of approaches to consider during implementation as well as focus the project on aspects that are critical to its success. For example, it is not worthwhile spending months developing a predictive model that is 95% accurate when an 85% accurate model that could have been developed in days would have solved the business problem. This time may be better devoted to other aspects that influence the ultimate success of the project. The accuracy of the model can often relate directly to the business objective. For example, a credit card company may be suffering due to customers moving their accounts to other companies. The company may have a business objective of reducing this turnover rate by 10%. They know that if they are able to identify a customer that is likely to abandon their services, they have an opportunity of targeting and retaining this customer. The company decides to build a prediction model to identify these customers. The level of accuracy of the ...