1. Asking the Right Questions
Asking the right questions before applying the appropriate techniques is paramount to getting the job done in data science. The approach to solving problems in data science is fundamental. It is a fact that if you are an inquisitive person, data science is for you, but you need to protect yourself from some eminent pitfalls.
Most people start working on the data from the wrong side. They take a data set and apply their most favorite tools. As a result, you end up with narrow questions like “Yes” or “No” and shallow arguments lacking the depth of knowledge and intellect, stating what is visible.
When you acquire data, you need to do ample thinking with a structure of data to avoid a futile path to simple questions and unsurprising outcomes. We need to avoid unsurprising results and focus on how much knowledge we can obtain from a given dataset.
2. Prioritize “Domain Knowledge” Over “Right Techniques”
Even as a professional data, one should always prioritize the domain knowledge over the techniques no matter how technically viable you are in your profession as a data scientist. Without domain knowledge, the efforts to perform manipulative operations that you perform on data fitting it into models to gain outputs by hook or crook are futile. Selecting the right techniques should come after the inquisition of the right questions. It is a mindset that is crucial to the success of any data science project. However, proficiency in both “domain knowledge” and the “selection of the right techniques” are highly desirable in the data science world, but now you know what to prioritize.
As a data scientist, if you want to create the information of lasting value, we have to reach an understanding of the following:
- Needs of our coworkers.
- The shape that work takes.
- Structure of arguments we created.
- The process of events after we “finish.”
To reach the above objectives, we should give ourselves the room to contemplate. We should take care of “why” and “what” before we entangle ourselves in questions beginning with “how.” Otherwise, our precious time goes to waste in taking the wrong actions.
A challenging part of dealing with data is to think about the “structure” rather than thinking in a vacuum. The benefits of having a solid structure in place prevent us from doing things that cross our minds. At the same time, the structure allows us to break down the problem and thoroughly study all the parts of the problem (also known as problem analysis). It makes this approach a fantastic problem analysis tool.
Human beings have been using structuring since time immemorial to ease the thinking about problems. We do not have to reinvent the wheel. We can modify ideas from other disciplines such as social sciences, English composition Philosophy, and Design to fit our needs and make our professional data work enormously valuable.
Creating a Scope for a Data Project
It is the first step once you understand the problem that needs resolution with your data science expertise. To find the structure in a problem, we first have to define the scope of the data problem. A scope is an “outline of a story” that revolves around a reason. The reason which we are working on, the actual problem and our expectations for the end of the story.
When we are working on a data science project, the project in a professional environment, it is likely to be an integral part of a more extensive setup. There may be people or teams that might get affected by the project or maybe are part of your team. A nicely laid out scope gives us a grip on the problem outlines and facilitates communications with stakeholders. There are 4 Parts of a Data Project Scope as follows:
- Context
- Needs (the project is trying to fulfill)
- Vision (of the achievement)
- Outcome
Finalization of Data Project Scope leads to conversations in the Data Science team, and the stakeholders become much more convenient, and thoughts can be written down. A convenient mnemonic for these five elements of data problem scope is CoNVO as in Context, Needs, Vision, and Outcome.
You should be able to hold a conversation with an intelligent, non-technical layman and a stranger, and he or she should be able to grasp the concept of the project on a high level. He or she should be able to understand the reasoning of accomplishments. In essence, no story is complete without a firm structure in place, and data project scopes are no different. The above described four elements of data project scope have applications to any structure of the story, and data storytelling is quite the same. One good piece of advice is that if you want to master scoping data problems, practice storytelling.
It is crucial to write down the CoNVO. Once we arrive at a clarifying piece of writing down to a few simple sentences, we can obtain data, clarify our understanding, and distill further to something smart and useful. In this matter, notice that data science is an iterative process.
1. Context (Co)
Context refers to the work that the people with whom we are working with and the work they are performing. Communication is the key to acquire the context of people, and a thorough understanding of their long-term goal is our primary objective. The context helps to establish guidelines to make a significant decision about the project involving data.
Context can be dynamic during a data project when new employees, partners, or supervisors join the organization or the mission of the organization abruptly transforms. Proper articulation of the goals of an organization is a vital part of gaining context.
2. Needs(N)
“Needs” are the things that are required to be fixed or understood to carry out the goals of an organization as every entity may face challenges sooner or later in its lifespan. The main objective of data science is to design steps to create knowledge. A need that can be satisfied with the power of data is in its pure essence about knowledge and knowhow. Understanding the mechanisms in the functioning of some part of the world is what need provides.
When we correctly lay down the needs in writing about what can improve with knowledge plugging all the empty wholes in our understanding, we are making progress in the right direction fulfilling the “needs” of our coworkers.
It encompasses the teachings gained by a spreadsheet, the information gained from a tool, anticipated information before making a previously unknown graph are all sources of “needs.”
3. Vision (V)
When a data science project launches, the first step is not gathering, collecting, or acquisition of data at all. So, the proceeding steps of performing transformations, testing ideas, and so on, are also out of the question. You have to envision the project first and ask some critical questions like where we are going and what it’ll be like to achieve our goals?
A Vision in a data project helps us to provide a glimpse of the obtained objectives and the ultimate goal. This glimpse could consist of a mockup, pinning down the expected results and a goal, an outline of argument that we are going to raise, and even some questions to narrow down our focus on our aims.
Coming up with a compelling vision as a part of the scoping process is most dependent on experience. It has its basis on the fact that the ideas that one carves out from the prior observations that one had during one’s lifetime.
4. Outcome (O)
At the last but not the least, before getting your hands dirty with data (collection), you have to consider this factor of the data project scope. The understanding of how the solution resonates back to change or even disrupt the organization as a data scientist.
Ask the following critical questions for a stable outcome.
- How is the solution supposed to be used?
- How will the solution be integrated into the organization?
- Who from the organization will perform this integration?
- Who is going to use this solution to make a difference in the organization?
- How will the success of the solution be measured?
The outcome is different from the vision. The vision focuses on the form of the work that is going to occur at the end. The outcome is the actual result or the solution. In other words, what happens when we finish. It is a critical data science practice that you adopt the best practices by realizing the true potential of a data project scope. On the same note, the previous understanding of how to prioritize your steps is crucial before you get your hands dirty with dirty data and succeed at every data science project that you undertake or signup.
