The Data Scientist
The Data Scientist extracts business insights from data, for a real world impact. They can spot unnecessary spending or fraud with Anomaly detection, find out proxy metrics for active users, Customer Lifetime Value, optimize ad spending to increase profitability, and generally, help the business with insights based on proven assumptions.
If a business embraces the data, there are lots of possibilities of reshaping the business based on it. A great Data Scientist ― when empowered by an organisation ― can come up with hypotheses to validate.
Data Scientists open the door. However, business leaders will have to enter themselves. ― Damian Mingle
Expectations and stakeholders
Like in the case of the AI expert, sometimes unrealistic expectations are placed on Data Scientists, which can lead to problems. Having the Data Scientist respond to multiple bosses, not empowering the team with enough decision power, or expecting overnight solutions to old problems are common problems. Other teams asking for adhoc reports and general data questions ― provenance and attribution questions are rarely easy to answer ― can pile up easily on the Data Scientist. These problems are compounded by the fact that KPIs of different departments depend on the data. Thus, the way this KPIs and metrics are calculated becoming a political issue in an organisation. In some organisations, the Data Scientist might find herself in the midle of these deciding issues.
The Data Scientist thrives deriving insights from data. But the data is rarely as clean as in data science competitions. The problem is often equally ill-defined.
Because of this, it is often the Data Scientist herself who is in charge of cleaning the data. This can quickly escalate into lots of tasks tangentially related to getting insights from the data. For instance, managing access credentials, backups.
Reporting is an important part of communicating clear data-driven metrics. Often, there is too many custom reports in ar organisation. When these metrics are constantly available to the whole organisation in easy-to-use dashboards, like Google data studio the whole organisation can function more efficiently. Custom dashboards can also work well, if the company has the technical skills to maintain them easily, and has a process to avoid too much customisation from stakeholders.
The time previously spent on single-use custom reports can now be spent in testing different hypotheses. For this, it is important that the dashboards are easy to use and change, and not a source of problems themselves.
Often, we can provide ― in addition to interactive dashboards ― programmatic access via SQL. This allows other technical departments to use the data from the Data Scientist.
To expand your skills
Here are some resources to expand your skills as a .