Soft expertise is an important part of the research of data science. Many non-technical skills are important for data science projects, from business value to interpersonal communication.
Communication
The ability to contact a non-technical public is a skill that each data scientist should have. The ability to contact a non-technical public is a skill that each data scientist should have.
“This is the most important skill, I think,” said Herman, who now is a professor of data science at The Flatiron School in New York City. “The most essential thing is to be capable of meeting the audience.”
Herman would concentrate on results when explaining his work to the rest of the company during his previous work on the railway. He would concentrate on the reasons for the model.
“Your model will not go into production if you can’t explain your model,” said Herman. “It would be only research I did for fun if I could not explain why this model would predict coal trains more precisely than we used to predict coal trains and nothing of the essence would come.”
But are not natural communicators capable of acquiring that skill? Absolutely.
In addition to technical writing, Herman’s students have to make a slide deck for a non-technical audience.
“They will present it to an instructor and the instructor is going to pretend that he is a non-technical person and ask questions about businesses,” he said.
Data scientists can practice outside of the classroom explaining their projects to family members or friends, meetings groups, and events.
With the pandemic, there have been a large number of meetings, technical conferences, and other groups where a data scientist may go to give a talk, but people still have opportunities to present themselves. And there are still presentations within companies, even though there’s no one in the office, Herman said.
“If I have created a model, it won’t be made unless I introduce it, even if this is Zoom type of meeting,” he said. He said.
Business focus
Data scientists often lack the ability in connecting theory and practice, says Bobby Rountree, the leading provider of technology services for federal agencies at Hitachi Vantara Federal.
“The value is to help customers get better ideas, invest better, and decide better,” he says.
Rountree acknowledges he didn’t start with a business mindset when he first began his career. Thankfully, this is a learning skill.
“This is how I could advance my career,” he said, “being able to surround myself with those who first thought about business.
In their industry, their businesses, and even the specific department or job function, data scientists must understand what makes sense. You must also be able to ask the correct questions, to know exactly what customers or users want.
“Sometimes they have no knowledge or change their minds about what they want,” he said. “It’s got a fly that you have to adjust.”
Data scientists may have to be researchers, AI Director Allen Hamilton, Kathleen Featheringham, said.
“We are asked to do the AI the number one thing.” She said, “Do what?” we asked them. “And it doesn’t mean you should just because you can.”
Once a business problem is identified by data scientists, the other side of the issue of business value consists of the use of the solution. There may be obstacles to adoption or another deployment, cultural or management, that can kill a data science project even if anything else is correct.
Fotheringham said, “People could revolt.” “You could think robots are coming to work. You must therefore look at the psychological aspects and ensure that you address the technical and human aspects. The worst thing is to create something no one can use.”
Another common problem is that when customers or users understand the scope of the problem, they do not take into account the key aspects of the workflow that they are used to doing so.
Another part of the business value of data science projects is often overlooked in the field of management. If an organization sends people to be trained in Python but the evaluation of performance is not changed and staff is not recompensed for using new skills, then the training will be wasted.
Data scientific experts can start with the idea of being a user, said Andrea Levy, data analytics director at Alation, a company of data.
She suggested, put on someone else’s figurative hat in the organization and ask what they want.
“Another way to better understand the picture is by working in an informal environment with other teams,” says Levy. She said, “Learn what they are doing, ask and generate data for them.
Ethics
When building predictive models, data scientists have a lot of power. They choose to select data sets, prioritize certain features over others and how data are used can influence project success – including the viability of an undertaking.
To perpetuate or exacerbate existing preconditions or to create new ones is one potential issue. Violations of privacy can result in bad relationships, compliance breaches, or a company’s failure just as it is being released.
It doesn’t mean it should simply because data can be collected. Sometimes the issues aren’t obvious right away.
“You might not be comfortable with forecasting gender-based loans if you’re working at a bank,” Herman said. “But another feature may be very gender-related.”
You have a role to play in explaining your actions. It not only facilitates communication of the value of a model to business participants but can also help people decide whether decisions are taken in accordance with ethical guidelines.
