In June of this year, the company Tail Target, in partnership with user groups, conducted a study on the profile of Java developers in Brazil. To do this, they shared a link among developers that was meant to be clicked by anyone wishing to participate in the survey.
This link did not contain a single question, nor did it lead to any questionnaire. Immediately, emails poured in warning that the shared link was incorrect. It wasn't. In less than a week, a complete study was ready, showing the interests, demographics, and lifestyle of the developers. Magic? No, Data Science.
Data Science is a new field that combines Big Data, statistical processing, and artificial intelligence to find information and detect patterns. It is increasingly common to find large companies whose decision-making is based on Data Science.
For others, Data Science is fundamental to the very existence of their business. On Netflix, 75% of the audience comes from its recommendation algorithm, which is a great example of applied Data Science. The relationship platform e-Harmony uses Data Science to find the ideal match for a person and is already responsible for 5% of marriages in the US.
The study on the Java developer profile analyzed anonymous browsing data from hundreds of people who visited one of the Java development websites being monitored. By analyzing which other websites these people visited, artificial intelligence algorithms detected patterns that allowed them to map out a behavioral profile of these individuals.
According to this study, 88% of Brazilian Java developers are men and only 12% are women. Adults represent the majority of these developers (44%), followed by young adults (30%) and teenagers (25%). Finance, technology, soccer, travel, and TV, in that order, are the subjects that interest developers the most.
The surprise in this study came when measuring the micro-segments that most interested male and female developers. Among men, the main micro-segments are free-to-air TV, soap operas, high-income cars, TV series, and international travel. Women, on the other hand, prefer to read about free-to-air TV, cable TV, hair, and makeup. In other words, male developers are soap opera fans. Female developers are not.
The applied algorithms do not collect demographic information or any information provided by users. What they do is process access logs from thousands of websites and run a series of artificial intelligence algorithms that attempt to guess demographic information and interests based on these people's online behavior.
Obviously, there is a margin of error in these algorithms, but they have become increasingly accurate. After the publication of this study, a researcher from the University of Kent sent data from a similar survey conducted in 2003. That study used traditional methods: interviews and questionnaires to find the demographics of the Java developer. The 2003 study identified that 88% of developers were men. This is exactly the same number found by the algorithms that conducted the same survey using behavioral data.
*Fabiane Nardon holds a PhD in Electronic Engineering from the Escola Politécnica of USP, a Master's degree in Computer Science from the Universidade Federal do Rio Grande do Sul, and a Bachelor's degree in Computer Science from the Universidade de Passo Fundo. Since 2012, she has been the Chief Scientist at Tail Target.
source