In my previous posts, I explained my view on Evolution of Data Science Architecture & The Transition to Data Science Architecture. In this post, I will explore data science and the role of the data scientist in an intelligent enterprise.
Albert Einstein once said, “If I had an hour to solve a problem, I’d spend 55 minutes defining the problem and only 5 minutes finding the solution.”
This idea is relevant today, especially in data science, where finding the right problem statement or use case is more important than finding the solution. Picking a language or technology is the easy part; it’s understanding the business problem that needs to be solved that’s difficult.
True data science is about understanding the customer’s business and adding value to that business by building data products on top of their existing software solutions. A good data scientist should enable this by communicating effectively with everyone across the organization.
As with any science, the best way to learn data science is by doing it. Data science should focus on following:
- 1Discover known unknowns to transform data into enterprise information assets.
- 2Enable predictive analytics to extract actionable insight from the data.
- 3Build modern data solutions and products to build confidence among business users, helping them make better business decisions and evolve into an intelligent enterprise.
- 4Assist in building data solution/platforms and products from any data form (structured, unstructured, digital, etc.) to align with consumer/customer preference and needs.
- 5Produce data ecosystems based on architectures that can reveal business stories from data.
- 6Enable IoT and digital business (including touch-points with mobile, social media, and cloud) by connecting all the devices by making information available anywhere/anytime.
The role of the data scientist
To analyze, secure, and process massive quantity of unstructured/digital data, we need data scientists with a mathematics/statistics background and knowledge of data analysis, data mining, data modeling, schema-less design, and data architecture.
To succeed, data scientists must also combine business knowledge with technical acumen, statistics, and mathematics knowledge. Today, intelligent enterprises are working to better align data scientists to the business by having them work with a variety of departments and business users.
Key data scientist attributes:
- 1A curious, skeptical, inquisitive mind
- 2An ability to translate technical language and clearly relate data science solutions
- 3The ability to act like a data artists, bringing creative, innovative thinking to data insight by slicing and dicing relevant business data or exploring data to discover unknowns
- 4An ability to explore the use cases that will have major impact on business for the customer.
- 5A highly trained professional familiar with business strategy, not a hobbyist
- 6The ability to build data products/solutions that deliver actionable insights
- 7A strong coder who is knowledgeable about data engineering but refrains from using technical jargon while communicating with business users
- 8A knowledge of model data, model algorithms like K-Means and Naïve Bayes classifiers, an understanding of data relationships, a knowledge of machine learning and predictive analysis
- 9The ability to clearly explain data visualization graphs and how they relate to the overall business
- 10The ability to explain why a solution was built the way it was built
In my opinion, data scientists should spend 25% of their time on business, 40% on data analysis, and 35% on algorithms/visualization/data science architecture. However, because data scientists must have such a wide range of skills, it is difficult to find truly excellent ones. That means some enterprises will need to coordinate multiple teams skilled in data engineering, traditional data architecture, business analysis, and business intelligence.
Now that we have a clear understanding of data science and role of the data scientist, the question is, “What can we use it to solve?” This topic will be covered in my next post.
Fecha de publicación: 22/10/2015