A veritable proliferation of job titles over the years in the analytics space, which this post will try to explain:-
Business Intelligence (BI)
After reporting, this is probably the most common term in the analytics space. It refers to people who write reports and dashboards which graphically demonstrate key performance indicators – KPIs (aka measures) for particular business units. Reports can be diced and sliced by dimensions which are grouped in to hierarchies and which have attributes. BI specialists will typically use tools such as Cognos, OBIEE, SSRS/AS, Microstrategy & Business Objects (amongst) others to produce reports & dashboards. The other main area of BI is data discovery/insights, where data can be pulled in to a tool and manipulated and graphed immediately in order that patterns/trends can be discovered more quickly than it takes to develop a report or dashboard. Tableau and Qlik are the best known tools for doing this.
Capture the BI requirements from the business, produce wireframes of reports and dashboards, and produce slide decks covering KPIs, Dimensions, Hierarchies, known data quality issues for approval by data governance bodies.
Profile data held at source to assist a data architect in developing a data model and a data engineer in cleaning the data prior to analysis.
Old job title for people with Maths PHDs who use statistics to identify patterns and control risk.
New job title for guys typically with Maths PHDs who use statistics to model patterns in the real world. For example, they might produce a model which calculates weightings to be applied to customer’s credit risk or potential fraudulent behaviour. Incoming data is then compared against the model and final scores computed.
Data scientists will typically produce models and visual output using programming languages with statistical functions such as R, Python, Scala and Matlab.
Also known as artifical intelligence. Neural networks used to be an earlier form of this area. It’s a subgroup of data science, where the model can be trained by using initial seed data, rather than having rules pre-programmed.
Refers to algorithms which detect hitherto unknown patterns in data sets.
Data scientists working on trading platforms for investment banks
Data scientists working in the insurance industry to price premiums and produce claims risk models.
Determination of likelihood of future data based on the known past. Usually comes with confidence ranges which get wider with time. Works well for deterministic problems but not problems with complex variables e.g. stock markets or global warming.
Term for roles which involve getting data from source to target without loss, cleaning, classifying and preparing the data so that analytics can work successfully.