Deconstructing a data science question is extremely important because you can generate business value in every step of a data science project. Our lead scientist Tessa Jones explains how to ask "molecular" and "elemental" questions to begin deconstructing a data science question to achieve the greatest ROI possible.
Deconstructing a Global Business Question
When beginning to deconstruct a data science question, you first need a global business question to answer. For this specific example, let's pretend you are a marketing person and you need to know how and when you should market particular products in your business. This question, by itself, cannot be answered because we need to know a lot more background information.
Questions like what the seasonality of your product is? Or, what is the future or past sales of your product need to be answered so you can answer your main global question.
In order to begin the deconstruction process, you need to ask molecular questions.
What are Molecular Questions?
Molecular questions in data science are intermediate questions that stand between your large global business questions and lower-level ones. When deconstructing your data science question, you may need to answer multiple molecular questions before you have enough information to generate a suffice answer.
Some examples of molecular questions you might ask include:
- What does our product demand look like?
- What will our sales look like?
- What is our customer segmentation?
These questions are defined from smaller elements that group together and form a molecule--just like science!
What are Elemental Questions?
Elemental questions are smaller, lower-level questions that you ask when trying to answer a molecular question. Elements supplement you in answering your overall global business question, as you need to spend time finding all the elements to define a molecular question in data science.
A few examples of elemental questions you might ask to define a molecular question include:
- What is the relationship between our products?
- What is our sales seasonality?
- Are there specific attributes that drive demand?
Connecting Elemental and Molecular Questions to Answer Business Questions
By asking these elemental and molecular questions, you not only solve your data science problem, you generate business value every step of the way. Connecting all of your elements and molecules does take some time, but it is extremely valuable for your business in the long run.
Each of the elements and molecules have value, so it is important to be as thorough as possible when deconstructing a data science question.
Conclusion
To conclude, breaking down data science questions into smaller ones can generate value in your business every step of the way.
Posted by Gage Peake