You can't get where you're going without a map. In data mining, that map is a process model. It provides a reliable framework to move from a raw business idea to a deployed, valuable solution. The most widely used framework is the Cross-Industry Standard Process for Data Mining, or CRISP-DM.

CRISP-DM breaks a project down into six phases: Business Understanding, Data Understanding, Data Preparation, Modelling, Evaluation, and Deployment. The arrows in the diagram show the most common paths, but the process is highly iterative. Findings in the Modelling phase might send you back to Data Preparation to create new variables, or even all the way back to Business Understanding to refine the project's goals. Let's start at the beginning.

The Business Understanding phase is the most important step, because it sets the direction for everything that follows. A project without a clear business goal is just a technical exercise. Your first task is to translate a vague business request into a concrete analytical goal.

Imagine a stakeholder says, "We need to sell more." That's a business objective, but it's not a data mining goal. To make it one, you need to ask clarifying questions:

• What are we selling? To whom?
• Are we trying to acquire new customers or sell more to existing ones?
• What factors do we believe influence a customer's decision to buy?
• What data do we have about our customers and their past purchases?

This new goal is specific, measurable, and directly points towards a data mining task: building a predictive model. Now you have a clear destination for your project.

Once you have a clear goal, you need to define what success looks like. This happens on two levels: business success and data mining success. Business success criteria measure the project's impact on the organisation. Data mining success criteria are technical measures that tell you how well your model is performing.

Defining these criteria up front is crucial. It ensures everyone agrees on the project's purpose and provides a benchmark for the Evaluation phase later on.

Finally, the Business Understanding phase involves a preliminary project plan. You'll assess the resources needed, identify potential risks, and check the overall feasibility. Is the necessary data available? Do you have the tools and expertise to complete the project? Answering these questions early prevents costly surprises down the road.

Remember, the goal isn't to create a rigid, unchangeable plan. The iterative nature of data mining means you will almost certainly revisit these initial assumptions. But by starting with a solid understanding of the business problem, you ensure your technical work is always focused on creating real-world value.

With a clear business goal and success criteria, you are ready to move on to the next phase: understanding the data itself.