News

Developer Economics Survey: Data Science, Machine Learning Are Most-Wanted Skills

• By David Ramel
• September 24, 2018

While perhaps not offering new conclusions about the hottest technologies in the software development arena, a new Developer Economics survey from SlashData provides more hard evidence that data science and machine learning are the top skills developers want to learn.

SlashData describes itself as an analyst for the developer economy, helping enterprises understand software developer audiences and measure the ROI of their developer strategies.

To that end, it recently conducted the 15th edition of its Developer Economics survey of more than 20,500 developers in 167 countries.

"Data science is the top skill to learn in 2019," SlashData said. It noted that 45 percent of developers want to gain expertise in data science and machine learning, with other most-wanted skills including UI design (33 percent) and cloud-native development (25 percent).

Other data- and analytics-relates skills also ranked highly, including data engineering skills such as ETL and data warehousing (22 percent) and marketing analytics (10 percent).

"Recent advances in deep learning -- including breakthroughs in Deep Reinforcement Learning (DRL) and Generative Adversarial Networks (GANs) -- have brought us many interesting applications of machine learning/AI in areas like autonomous driving, computer vision, natural language processing and medical diagnostics," SlashData said.

"The analysis of very large datasets is now made possible and, more importantly, affordable to most due to the emergence of cloud computing, open-source data science frameworks and Machine Learning as a Service (MLaaS) platforms. As a result, the interest of the developer community in the field is growing steadily. The three most popular use cases in the past year have been image classification, customer behavior analytics and natural language processing, targeted by at least 20 percent of machine learning developers and data scientists. However, the majority of them (26 percent) are still exploring their options and have not decided yet in which area to specialize."

Other data science- and AI-related findings included:

• A small minority of data scientists work on large volumes of data, and only 21 percent who generate real-time predictions. To add to the "small data" case, the majority of those generating real-time predictions (68 percent) produce no more than 10,000 real time predictions per month -- and the numbers are similar for those generating batch predictions.
• It is in the much-discussed areas of speech recognition, image classification and natural language processing (NLP) that more than 50 percent of data scientists are using small training datasets of no more than 20,000 records.

The firm offered up the following additional key takeaways from the survey, which was conducted online over seven weeks between May and June of 2018.

• JavaScript is still the most popular programming language, with over 10 million users. Also, 3 million developers joined the community in one year.
• Python has reached 7 million active developers and is closing in on Java in terms of popularity, thanks to 62 percent of machine learning developers and data scientists who now use Python.
• DevOps has entered the mainstream. One eighth of the developer population is working on DevOps projects.
• There is lots of interest in robotics from developers but limited activity. 40 percent of developers expressed interest in robotics but only 9 percent of them were engaged in actual projects.
• Game developers are making more money. In the first half of 2017 only 29 percent of game developers were making more than $100 a month jumping to 48 percent in the first half of 2018.
• Individual developers seem to have less influence, not more, when as a group they are the majority in a company, or even a sizeable part of it. It seems that when developers are a relatively small group in a company, they get to take their own decisions on their tooling. If they are a large group, a hierarchical structure emerges within the development team and decisions are concentrated at the top.
• The smaller an organization is, the more likely a developer within that organization is to be involved in the purchasing process. However the percentage of developers without any influence does not go above 40 percent for any company size.

In addition to shedding light on the top skills developers want to learn, what areas developers are working on, learning about or interested in, and programming language communities, the survey report focused on four other major areas, including: the evolution of business models in gaming in the past 12 months and discuss the driving forces behind it; developers as decision-makers; and Big Data and real-time predictions.