Israel Innovation Authority to help train data scientists

Michael Li

Michael Li, founder and president of The Data Incubator, talks about the need for more data scientists worlwide and in Israel.

"In recent years, data science is taking over the world - not just in technology, but in fields like health, financial services, retailing, natural resources, and others. The challenges facing Facebook and Google in trying to enter the sector pale in comparison with the challenges facing non-digital conventional companies," US data scientist Michael Li, founder and president of The Data Incubator, which operates a program for training data scientists in eight weeks, tells "Globes." Li visited Israel recently as a guest of the Israel Innovation Authority in the framework of Artificial Intelligence Week held at Tel Aviv University in cooperation with Intel and the Innovation Authority.

Use of data science has been growing in recent years because there are now more data, following the connecting of many devices to the Internet and the realization by organizations that important business insights can derived from this data. This in turn has increased the demand for data scientists, who use machine learning tools in order to produce these insights and make forecasts.

The existing need in organizations and the realization that those who do not employ them are liable to be left behind has resulted in a shortage in the market. According to current data from technology placement company Ethosia, Israel lacks nearly 400 professional data scientists. Li's training program is designed to help meet this need. Li says that his academic background - a BS in computer science from Princeton University, an MSc in mathematics and statistics from Cambridge University, and a doctorate in applied mathematics from Princeton University - led him to the data science field. After finishing his studies, he worked on Wall Street as a quantitative analyst at JP Morgan, DE Shaw, and Bloomberg. "The job is similar to a data scientist, but it was before they called it that. The technological tools weren't as good as they are now, so to make the same analysis was more expensive and required better skills. The only place where it was profitable to do it was therefore in trading on the capital market," he says.

Afterwards, he became the first data scientist at the Andreessen Horowitz venture capital fund, from where he moved to Foursquare, one of the fund's portfolio companies, which devised an app for recommending places close to the user's location. He founded The Data Incubator in 2014 out of a desire to help academics who wanted to work in industry, and to help managers with their employees.

"We are now capable of gathering large quantities of data, because we have smartphones, laptops, wearable devices, and all those devices send updates to the cloud in real time. We're living in a very measured world, and with all of this information comes a huge computational challenge. Data scientists have to learn techniques in order to work with gigabytes of information, if not terabytes and petabytes," Li says. "There's also a lot of value in this quantity of information, because with more information, you can train much better models."

Li emphasizes a number of other challenges facing companies that integrate data science in the organization. "Companies that begin this journey have to go through a number of maturity curves: an analytic curve, which is the ability to take a lot of information and draw conclusions from it; an operational curve - the ability to take everything that we learn from the analysis and apply it to the real world; and the organization curve - bringing the entire organization to a situation in which it can adopt the changes derived from the information," he explains.

According to Li, despite the difficulties faced by conventional companies entering the field, it is important for them to adopt data science and artificial intelligence, for two reasons. The first is consumer demand: "Today, consumers expect every company to provide them with a consumer experience as simple as buying from Apple's App Store or sending a message on Gmail, and an immediate and intelligent digital response. It's forcing companies to use more analytical tools, data science, and artificial intelligence," he says.

The second reason is the competitive dynamic. "For example, take the recommendations engine. Consumers demand a lot of content, but they don't always know what they want, so you have to help them find content. One of the things that enabled Netflix to be better than other video providers was that it had the data that enabled it to do this," Li comments.

A manager cannot afford not to understand the field

In order to train data scientists and deal with the shortage of professionals in this field, Li describes two courses in The Data Incubator. The first is the Fellowship program, which lasts for eight intensive weeks and is designed for people with a PhD, or at least an MA. "The idea is to take every brilliant people who can learn the subject quickly and help then fit into the industry, thereby increasing the supply of experts," Li explains. After the training, The Data Incubator also helps its graduates get hired at companies that pay for recruitment, thereby enabling the organization to offer the training for free. The program operates in four locations in the US. Every quarter, 3,000 candidates file applications, only 2% of whom are accepted.

Li says that the students in the program can come from diverse disciplines, even from the social sciences. "When people think about a data scientist, the typical profile is somebody with a background in physics, mathematics, or maybe computer science. What we found through our rigorous selection process, however, is that people with a background in economics, chemistry, and neuroscience, for example, can also sometimes be very good data scientists. At each stage of the selection process, we assess different talents and techniques: writing code and programming, mathematics and statistics, and even the ability to make presentations. We're looking for people with the powerful combination of talents necessary to become data scientists," he says. According to Ethosia's figures from a year ago, 75% of the data scientists in Israel studied mechanical engineering, electrical engineering, industry and management, information systems, or the exact sciences.

Another course offered in The Data Incubator is for people currently working in corporations. "We have worked with dozens of Fortune 500 companies, and with the US government, the US army, and the UK government, and helped them train existing work teams. Adapting and defining courses of study is done with the companies according to the employees' current level in the field and the level that they want to attain. Another course we recently began offering concerns the business side of data science. The course, which is for two days, is designed for managers who don't have technical backgrounds, but who must understand what data science is, how to work effectively with data scientists, the subject of privacy and the legal consequences of using data, etc. According to a study by McKinsey & Co., there is a lack of 140,000-190,000 data scientists worldwide, but also of ten times as many data-savvy managers," Li says.

Li was visiting Israel to offering similar training to organizations in Israel in cooperation with the Israel Innovation Authority as part of its new program called "The Workshop for Advanced Technology Training." The new program will offer training for employees in various technology sectors, with the initial focus being on data science and artificial intelligence. Israel Innovation Authority social challenges division head Naomi Krieger Carmy says, "Every Israeli technology company now needs a data scientist, and you can't simply wait for them to arrive with a university doctorate. It's important for us to enlarge the pie, because companies today are stealing the same people from each other.

"We're therefore taking an interest in models like the training that Michael Li offers, which can take people from different backgrounds and help them in a fairly short time. The new program enables a group of five technology companies or more to join together and offer a training program for their employees. They can use an external organization, such as The Data Incubator, that offers this training." Companies can submit training proposals to the Innovation Authority until February 9, and the Innovation Authority will help fund up to two thirds of the training cost. "We'll succeed in enlarging the pie when we begin thinking how companies can cooperate, instead of competing," Krieger Carmy says.

Krieger Carmy also commented on the question of management, and the need to train managers lacking technical background, "It's just as impossible as to imagine a CEO who doesn't understand finances. He can't just say, 'I've got a CFO who deals with this; I don't deal with this part of the business.' We understand that data will be similar to this in many ways. A manager cannot afford not to understand the field," she says.

You also have to know how to tell a story

Li emphasizes that The Data Incubator's program differs from higher education study programs: "We're much more connected to industry than institutions of higher education, so we manage to stay abreast of the most innovative trends. While academic studies are very theoretical, our goal is to make the participants in the program productive as fast as possible. The theoretical part is less important for employers, but the people managing higher education institutions are professors. They get promoted on the basis of their research, and their research involves mainly what is theoretically interesting, in contrast to solutions that simply contribute making things happen and help companies make money."

As part of the focus on the practical side of data science, part of the study program concerns communications skills. "I think that one of the most important things in data science is the ability to persuade the manager to act in a certain way as a result of the insights that you derived from the analysis that you did. For this, you have to know how to simplify things and tell a story," Li says.

"Globes": What do you think about the tools and courses that enable people to learn the subject independently?

Li: "Many of the people who come to our program have spent time in online courses. In my opinion, this is a great place to begin. In our experience, most people need to be led by the hand a little, because in machine learning and artificial intelligence, you can get stuck in places where someone with experience can see clearly why you got stuck. A guide can help you escape the impasse and save days and weeks in the learning process. Especially in the organizational aspect, in which time is money, it's very important to organizations that their employees learn the material as quickly as possible."

Do you think that education should change according to the new technologies even before higher education institutions?

"We have to change the educational paradigm for education in junior and senior high school, so that it's less focused on differential and integral calculus and more on linear algebra and programming. The focus on differential and integral calculus is a heritage of the Cold War, when the most important thing was to build more nuclear missiles. That's obviously no longer important; we have all the nuclear missiles that we'll ever need. What we really need to do is research in fields like biomechanics, biology, and artificial intelligence. These are the new industries of the future, and this means that we should change our emphasis in mathematics."

Published by Globes, Israel business news - en.globes.co.il - on January 14, 2020

© Copyright of Globes Publisher Itonut (1983) Ltd. 2020

Twitter Facebook Linkedin RSS Newsletters גלובס Israel Business Conference 2018