How Data Scientists Spend Their Time

data_scientist

As you draft a job description for your new data scientist, you probably have a good idea of what you want them to accomplish. But how will they spend their time? Whether you’re starting fresh from scratch or updating a pre-existing document, you need to provide candidates with a clear idea of the tasks and responsibilities they’ll face if they get the job.

In its 2015 Data Science Salary Survey, O’Reilly Media not only asked data scientists about their compensation, but also about how much time they spent on specific analytics tasks during their workday. The online survey was self-selecting – it was promoted to the O’Reilly audience but the link was available to anyone – and it received 820 respondents globally.

The survey’s findings provide insight that can help you focus on the job description’s most important responsibilities. They can also be used as benchmarks for evaluating the composition of your team. After all, having the right mix of talent is crucial for accelerating your analytics efforts.

Here are 11 takeaways from the survey.

Data Scientists Spend Most Time on Basic Exploratory Analysis.

  1. Length of the workweek: 75% of respondents work 40-50 hours per week.
  2. Time in meetings: Meetings are a staple of the workplace, so it’s not surprising that 50% of data scientists spend at least one hour a week participating in them. For a notable 12%, meetings consume at least half of every day.
  3. Basic exploratory data analysis. The largest share of the week is devoted to this activity, with 46% spending 1-3 hours each day on it, and 12% spending more than four hours per day.
  4. Data cleaning comes in second. Forty-two percent of analysts devote 1-4 hours per week on this, and 31% spend 1-3 hours each day.
  5. Machine learning and statistics. A majority – 63% – of data scientists spend less than four hours per week on this activity.
  6. Extract, Transform and Load (ETL). Close to half of the respondents reported spending less than than an hour per week on ETL.
  7. Data visualizations. Nearly 80% of spend at least one hour per week transforming data into easily understandable insights.
  8. Presenting analysis. Nearly half spend 1-4 hours per week presenting analysis, with 6% spending four hours or more per day sharing findings with management.

Because this was a salary survey, it also provides insight into the relative value of each task.

The more time spent in meetings, the higher the pay.

  1. Widest salary ranges: Respondents who spend more than four hours per day on ETL, machine learning and statistics, or presenting analysis have the widest range between the lowest and highest salary.
  2. Meeting time impacts salary. The more time data scientists spend in meetings, the more they earn.
  3. Highest starting pay range. Only 5% of respondents spent more than four hours per day on ETL. However, this group also reported the highest starting pay, at just below $90,000.

To perform to the highest standards, your team will need to include data scientists working on each of these tasks. When putting together the job description for your next hire, these insights can help you identify gaps and define the new role accordingly.

FacebookTwitterEmailLinkedIn