Why technical knowledge isn’t the only thing you should pay attention to.
Data scientists tend to focus on the hard skills. They enjoy to learn about new topics and want to discover the latest developments in the field. And of course, it’s more fun to play with DALL-E 2 than to spend time on a presentation. Sometimes it’s easy to forget about the soft skills, while they are an essential part of being a good data scientist.
This post explains three essential soft skills for data scientists. If you practice them, they will help you, save time and make you stand out. Let’s start.
Note: I like to make the distinction between research data scientists, who focus on doing AI research, and practical data scientists, who develop AI applications for companies. The three skills are useful for both types of data scientists, although this article was written with practical data scientists in mind.
Many stakeholders expect revolutionary changes from data science projects. Don’t we all know an over-enthusiastic manager who wants to replace complete processes with AI? Or the CEO who is promised billions if he only would add AI to the company? After hiring a bunch of data scientists and waiting for months (or years), these people get disappointed. Why isn’t the company or process changed completely? They thought AI would give them golden mountains and it hasn’t. It’s hard to implement and everybody continues doing the same things as before (in the worst case).
“AI is overestimated in the short run, but underestimated in the long run.” — Amara’s Law (edit)
To prevent the disappointment from happening you can do several things. Besides starting well designed proof of concepts, in which the problem is clear and demarcated, you need to manage the expectations of your stakeholders. You need to explain AI isn’t magic and can’t change messy data or a bad infrastructure. You can explain dependencies that make it harder to make the solution work in production. Organize a meeting with stakeholders every one or two weeks in which you explain progress and talk about setbacks. Ask for help if you can’t solve the problems. Say no when you believe AI isn’t the best way to solve a certain issue and explain why. Keeping the process transparent and clear will help you and the company.
There’s also the opposite side, where people are critical towards AI. They might think: ‘Will AI replace me in the future?’ or ‘I don’t understand it, and that makes it dangerous.’ In this case managing expectations is also important. In most cases AI will not replace them, but it will replace repetitive, boring tasks. They get more time to be creative and discover opportunities for the company. This message can change the way you work with these people.
There are many reasons you need to have presentation skills. For instance, you want to convince people of your AI solution for a use case. It’s easier to convince them when you explain in plain and simple language why your solution needs to be developed. Or if you made a successful AI application, people will ask you to explain how it works and to present it. Presentation skills can also come in handy when applying for a job.
It might be hard at first, if you don’t have experience with presentations and you don’t like to be in the center of attention. Making hard concepts easier can also be difficult. There’s no golden rule, but an important thing you should keep in mind, is: what does my audience need to know and what’s their competence level? If you need to speak in front of the whole company, where many people don’t have a data science background, keep it simple and make it fun (add practical examples, remove text and formulas and add images or animations). If you have a small specific audience, explain a bit more about what’s in their interest (don’t bore sales people with math). People can always ask questions if they want to know more. Or you can set up an in depth meeting with a smaller group.
Where should you start to learn this skill? If you want to explain hard concepts in an easy way, you can search on YouTube or Google to see how others explain it. Or you can make an easy example from practice that’s understandable by everyone. An intuitive level of understanding is enough in many cases. A source I can recommend if you want to create easy understandable graphs is the book storytelling with data.
“Make it as easy as possible for your audience.”
This skill is twofold. On the one hand, curiosity is important so that you stay informed of the latest developments in artificial intelligence and to keep improving yourself. On the other hand, curiosity is important to get all the information you need to start a project well (or during the project). This section focuses on the latter.
If you are a bit like me, you love to solve problems. When anyone comes to you and asks: ‘Can you help me with this?’ you jump in and start solving it. After a couple of weeks, you check in with the person and you show your solution. He says: ‘That’s not exactly what I need!’ and you go back and need to redo a lot of the project. It is a pitfall to start too soon and thereby miss important information. You need to be curious to get to the core of the problem.
This skill is essential because data scientists have to deal with large and difficult problems. What is the underlying question of an assignment? When is the project successful? What is the baseline? What is done in the past to solve the problem and where are the results of those experiments? Are you the right person to carry out the project? What is the impact of an incorrect prediction? How can we make the project work in production? How does the solution connect to existing systems? Asking good questions will help you and save you a lot of time. You won’t be faced with unexpected surprises and are able to check your assumptions. The problem and constraints will be clarified and you understand the business motivations and technical requirements.
If you find it difficult to come up with the right questions, don’t worry. There are tools that can help you with this. For example, you can fill out a use case canvas, which allows you to monitor all aspects of a problem. Here and here are some examples.
“Starting a project well is all about asking the right questions.”
These skills have something in common: they are all about communication. Some lucky ones can give beautiful presentations, ask spot on questions and explain in a way that everyone understands, but most people need to practice and learn these skills in the field. If you want to do a course, a product owner course can help with managing expectations. And there are many presentation courses available. But the best way to learn is to start doing: Don’t hesitate to give a presentation the next time someone asks you to!
Good luck, and if you got questions, additions or anything else, please reach out!
Three Essential Soft Skills for Practical Data Scientists Republished from Source https://towardsdatascience.com/three-essential-soft-skills-for-practical-data-scientists-e73ce99f7018?source=rss----7f60cf5620c9---4 via https://towardsdatascience.com/feed