In 2015 67% of data scientists were happy or very happy in their job. What has happened in the last two years? 88% of all data scientists are now happy or very happy in their job. Interested in why this increase, or whether you should pursue a career in this area? CrowdFlower gives us lots of interesting statistics in their 2017 Data Scientist Report.
New Data Scientists
In 2015 25% of data scientists had been in their roles for less than two years. Two years later, that number has increased to 35%, a clear indication of many new data science graduates from the 551 colleges worldwide offering degrees in data science.
64% of data scientists agree that they are working in this century’s sexiest job.
Nearly 90% of data scientists are contacted at least once a month for new job opportunities with over 50% contacted on a weekly basis and 30% being contacted several times a week. They are in demand!
Modeling and Algorithms
Data scientists enjoy building and modeling data, mining the data for patterns and refining algorithms. Data scientists spend 38% of their time doing these things they most enjoy.
The bad news? 51% of a data scientist’s time is spent collecting, labeling, cleaning and organizing data. A whopping 60% of data scientist’s list ‘cleaning and organizing data’ as one of their least favorite tasks. 51% complain about ‘labeling data’ and 48% put ‘collecting datasets’ as one of their top three most dreaded ways to spend time.
While there’s no shortage of data, access to quality data is definitely an issue. Specifically when it comes to AI projects, 51% of respondents listed issued related to quality data as the biggest bottleneck to successfully completing projects.
Where is unstructured data coming from? 91% text, 33% images, 15% video, 11% audio and 20% other. 41% of this data is coming from public data sets, 41% is sourced by their team or through their own efforts but the vast majority, 78%, is generated by internal systems.
Beside their day-to-day work, what worries data Scientists? The biggest concern by 63% of data scientists is the problem of programming human bias or prejudice into machine learning. 49% are concerned about the implications/issues of utilizing AI and automation in warfare and intelligence and displacing the human workforce with machines concerns 41% of data scientists.
CrowdFlower summarizes that if 2016 was the year of the algorithm then 2017 is the year of training data. The integrity of this data is the “key to providing unbiased models as AI starts to drive our future.”