Data science process: 1. Defining Research Goals

Data science process is a series of activities that help us move from a business problem to a useful solution based on data. 1. Define research goals 2. Data retrieval 3. Data preparation 4. Data exploration 5. Model building 6. Presenting results and automation

Defining Research Goals

• Before performing data science project, we must first understand what the project is about, why it is needed and how it will be done.

• Every project begins by answering three important questions:‒

1) What: What is the goal of the project?

2) Why: Why is the project important for the company?

3) How: How will the project be carried out.

For example ‒ i) what is the project? "predict customer churn" (who will stop using the service). ii) Why to predict customer churn ‒ because we want to reduce the customer loss and increase profits. iii) How to carry out this? ‒ By collecting customer data and building prediction model.

•  The goal of asking these questions is to make sure that everyone ‒ team members, clients and management clearly understand the project's purpose and agrees on what needs to be done.

1. Understand Goals and the Context of the Research

• Finding the goal is the main purpose of your project ‒ what exactly you are trying to find or prove.

• Spend time talking to managers or clients to understand their expectations. Ask clear questions ‒ What, Why and How.

• Keep refining the goal until it is specific and measurable. For example ‒ Research Goal : "To develop a model that predicts which customers are likely to cancel their subscription in the next 3 months." This goal is clear, focused and practical.

• Many data scientists fail not because of weak technical skills, but because they don't truly understand what the business needs. Hence understanding context matters a lot.

• Defining research goals ensures you know what to do and why you are doing it.

2. Create Project Charter

• Once you fully understand the business problem and research goal, the next step is to create a Project Charter.

• A Project Charter is a short document that summarizes all key information about the project.

• It acts as an agreement between the data science team and the client or management.

• A well‒prepared project charter should include the following:

■  Clear research goal ‒ What you are trying to achieve.

■  Project mission and context ‒ Why the project is important.

■  Methodology ‒ How you plan to perform your analysis (tools, techniques, algorithms)

■  Resources needed ‒ Data, software, people and hardware required.

■  Feasibility proof ‒ Evidence that the project can actually be completed (a small test or "proof of concept").

■  Deliverables ‒ What you will produce (for example, a report, model, or dashboard).

■  Success criteria ‒ How you'll measure if the project succeeded (accuracy %, improved sales, etc.)

■  Timeline ‒ Key milestones and completion date.

Sample project charter

Component     ‒     Example for a customer Churn project

Goal ‒ Predict which customers may cancel their service.

Context ‒ Company wants to reduce the customer loss.

Method ‒ Use Python to build prediction model using customer data.

Resources ‒ Customer Data, data scientist and data analyst.

Deliverables ‒ A working prediction model with dashboard.

Success criteria ‒ Model accuracy must be above 85%.

Timeline ‒ Two months.

• Creating a project charter ensures everyone agrees on how it will be done and what success looks like.

Review Questions

1. Explain how will you define the research goals during the data analysis process?

2.What is project charter? Explain.