A Bug That Can Make You a Data Science Hero
What if I tell you that there is a bug that can take you on a ride in the world of data science. Yes, if you have the bug of curiosity, consider yourself the best fit for the data science profession.
A Useful Bug?!?
"Bug" is typically not a positive word, and is something we like to avoid in general. Ask software developers and they would dread bugs.
Source: Web vector created by storyset - www.freepik.com
But what if I tell you that there is a bug that can take you on a ride in the world of data science. Yes, if you have the bug of curiosity, consider yourself the best fit for the data science profession.
If you do not take the things as given and have a knack for learning how everything is done currently, chances are you will end up building something that will be better than what exists now.
Not everyone has the inquisitiveness of asking the ‘What’s, Why’s, and How’s, without the fear of being judged. We all are molded by the assumptions that everyone must already have thought of what I am thinking of.
"What if I ask questions and I get labeled as unqualified, inept, inexpert, and uncompetitive? What if I am not rightly skilled to deliver the appropriate data science solution"
These feelings have a name: Imposter Syndrome.
It exists in every field, but it is more ubiquitous in the data science world. The imposter has its own reasons to persist considering the breadth and depth of the machine learning concepts and algorithms.
You are expected to wear a magician cap and run the magic wand to churn some model that will always remain truthful to its developer, the one who gave birth to it.
But that never happens. As against what is thought of a data scientist, the magic (if it all exists) comes from the data itself. You can’t take any dataset and start building a model recklessly.
The guy who said “there is no free lunch” was not joking. It is for real – every business problem is different and so is its objective, data, its characteristics, and lifecycle.
But back to our original point on curiosity bug, the snooping data scientist is immune to such sham feelings and is beyond the thoughts of “What if I am not”.
The curious data scientist will seize every opportunity to leave a footprint by undertaking the given problem.
It Isn’t That Easy
Having an innate skill of being inquisitive around you – sounds like an easy trait to develop. But trust me it is not.
Like any other power when exercised improperly yields inferior returns, the same holds true here. You cannot be negligent of what most folks are discussing in the forum and ask repetitive questions.
To challenge the status quo i.e., the current state of the solution, you need to show the scope and perform the gap analysis. This is called sizing the opportunity.
You cannot lay your finger on anything and start working on it. It needs to have a scope of further improvement in the first place.
Low Hanging Fruit
It is relatively easy to cover the journey of 0 to 70% by building scratch up and generating huge returns. But once a baseline is generated, building marginal improvements on it is not that easy.
You may list ‘N’ number of things with potential scope of improvement, but it is not in your hands solely. The business must onboard your suggestions and will decide which problem has a better ROI of time and resources and can be considered as quick wins.
As you cannot work on all business problems at once, you need to consult with the business on the priority list of what adds more value.
Watch Out Before You Commit
Data Science is all about finding business problems and building data-driven solutions. To start with, data needs to be in place and must have patterns to learn from.
Pattern recognition, by definition, includes the use of machine learning, aided with the increased availability of big data and a new abundance of processing power.
Note that once you are on board, nothing is set in stone. You can very well find yourself working on automation jobs, designing simple heuristics to get those early business gains, though no one will explicitly tell you as part of the job description.
So be ready to embrace the fact that not all data scientists get to do all ‘premium’ modeling work all the time.
Time For A Reality Check
Source: Background vector created by redgreystock - www.freepik.com
Surreptitiously speaking, they are responsible for all things related to data that can achieve business objectives.
Business in general is not particular about which algorithm to use and how you solve it. They expect you to understand their pain points and help use data to overcome them. It is your prime responsibility to share with the executives and stakeholders what problems ML can solve and where it would be only pieced as aesthetics.
Ask questions to understand how your solution can add value. Identify and explain the pros and cons of using ML to solve a problem.
In short, if you focus on the end goal of lifting business numbers, all the tools and techniques used become your ally.
Foster That Childlike Curiosity
Children assume no prejudice in asking questions – relevant or irrelevant. Though we professionals cannot afford to ask irrelevant questions out in the open. But who knows and decides what is relevant and what’s not? If you need that information to deliver a solution – ask it. Data Science is iterative and needs to discover the “Whys” quite often.
No one person is sufficient to know it all and deliver single-handedly, hence it calls for stout teamwork. The journey becomes easier if all can brainstorm together and identify the action items.
But you cannot wait for the rest of the world to be on the same page with you, hence it becomes your responsibility to wear that curiosity hat and become the hero of your data science solution.
Vidhi Chugh is an award-winning AI/ML innovation leader and an AI Ethicist. She works at the intersection of data science, product, and research to deliver business value and insights. She is an advocate for data-centric science and a leading expert in data governance with a vision to build trustworthy AI solutions.