A Data Science Central Community
While everyone talks about Data Scientists and there is extensive discussion on who a Data Scientist is, I've experienced one factor being overlooked while creating an analytics project. Essentially, that the Data Scientist isn't a person... it is rather a team.
Now, I have been busy with Sensor Data analytics for the last few months. I might have missed reading up what some people have been talking about on this aspect. However, what I am writing here is more from specific experience rather than general observations. The following roles could be thought of as key ingredients for for a successful analytics project:
1. Columbus (The Business Analyst + Project Manager) - This is the keystone in the arch. An explorer conversant with business aspects as well as data science terms and concepts. This is the person who can lead your team to find new solutions, dream about new applications and avoid being overly mathematical when talking to users.
2. Helmsman (The Logistics Manager) - Initial phases of projects need a lot of background logistics, travels, workshops and calls. A competent man at the helm can carefully steer the analytics project through data access and privacy challenges, organizational skepticism and help the analytics team focus clearly on the goals at hand. This role is often ignored and Columbus is put upon a very hard task... to plan, dream and chalk the path and even steer the ship in the process.
3. The Spanish Monarch (The Sponsor & the SPOC) - These people from the client organization work with Columbus and the Helmsman to iron out challenges from the client end and ensure smooth operation of the project.
3. Pythagoras (The Advanced Data Scientist) - A more mathematically oriented data scientist who not only looks at utilizing techniques but also on building algorithms for new techniques.According to them, everything in the world... even things like music and art can be expressed in mathematical terms. You need to decide beforehand if your project needs an extreme data scientist since these people's skills might be wasted if all you need is some basic regressions and classifications. Also, these people would rather work on neural networks and SVM rather than build simple least squares regressions. These experts are great if put up the task of exploring a completely new area or for developing techniques that would take the world ahead beyond the company's business.
4. Zuckerberg (The Intermediate Data Scientist) - Typically, a these people classify everyone into 2 categories - Those who understand Mathematics and those who are stupid. (A blatant exaggeration... I agree. Just highlighting a point.) If you are looking at solving a problem where some previous guideline exists or in a common field such as Market Basket Analysis, these people are great. However, they sometimes have a tendency to contradict the Helmsman.
5. Good Will Hunting (The Gifted Novice Data Scientist) - Some people have a knack for Data Science just as some have a knack for Mathematics. These are rare to find and motivate. However, if channelized perfectly, these can work wonders for projects.
6. The Apprentice (The Novice Data Scientists) - A novice with some years of experience in solving data science problems. Can also be tasked with the important time-consuming task of gathering and sorting the data. Validations and checking could also be passed on to these.
7. Picasso (Domain/ Business Expert) - A person who understands the intricacies of the applicable business area. This person could either be involved with the rest of the team or could be an external consultant to provide insight and direction. Typically, this person comes from the academia or from the client organization.
Besides these, depending on the size of the project, the following could also be included in the team:
8. The Cartographer (The Data Architect)
9. The Cheif Officer (The Data Collector/ DBA/ Data Cleaning Expert)
10. The Cook (Data Expert) - Sometimes enough data is not available and simulations might be required.