Data analysis means different things to different people:
- To a business manager it might mean enabling and encouraging data-driven decision-making.
- To a newspaper it could involve collecting data on a particular story and incorporating it into an article, probably including some kind of visualisation.
- To an online marketing team data analysis may be the website analytics that help inform how they divide their budget between different channels.
- To a UK rail franchise bid director, it is a regulatory requirement to forecast what impact their actions will have on trains’ punctuality and how much delivering the service will cost.
Data analysis has evolved massively in recent years with developments in data storage and processing that have led to the term big data being coined. However, the fundamentals have not changed and 50+ years: all data science projects necessarily begin with a problem or question and a key aspect of the data analyst’s role is to help correctly define this question before carrying out any analysis. A quote attributed to John Tukey is helpful to keep in mind:
Far better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise.
As such, selecting an analyst with whom you can communicate easily and clearly is as important as finding one who is technically proficient.
The “What”: Your business questions answered by leveraging the power of varied, relevant and carefully selected data sources through data processing, statistical analysis, visualisation and inspirational communication.
The “How”: As with any kind of technical consulting service, data science begins with understanding the client’s business and the problem to be solved. By definition we are talking about business problems for which a quantitative rather than a qualitative solution is felt to be most appropriate.
We answer questions like:
“How much revenue will option A generate compared to option B?”
As opposed to questions like:
“What will the relative impact of options A and B be on our company’s reputation?”
A typical data science project lifecycle will look something like the one shown below from EMC:
Similar to the diagram shown, the service I offer focuses on delivering data-driven insight to answer a specific business question. An additional step often included in data science and business intelligence lifecycles is the implementation of a software solution by which the insight becomes available to the business whenever the same question arises. This is outside my area of expertise.