Data Science/ Transition

Hi, I am having 10+ years of experience in System Administration(LINUX/UNIX) and i am interested in Data Science. Is it right choice to get transformed to data science and what are all fields i can try?

Hi

Data science is an intersection of various fields like structured problem solving, computer science (specially optimizations and algorithms) and mathematics.

Typical work of a data scientist is:

  1. Meeting with stakeholders to understand requirements
  2. Create clear business goals of WHY we need to develop this prediction system/ data science model. What metric are we chasing?
  3. Deciding which features impact the metric the most
  4. Getting data and cleaning it to get it in a usable format. Also implementation of features in code. This takes like 50-60% of the entire time
  5. Deciding the target variable and creating the model. This take 10% time but looks impressive in 90% presentations and is kind of the end result
  6. Tuning the model
  7. Deploying it in A/B testing environment and iterating over the model again and again to improve it further. Whether identifying segments or doing multiple models on them
    Typical tools used here are Python, R, SQL, Mongo etc

Big data analytics is generally more focused on making machine learning and data science models scalable. Typical technologies here are Apache hadoop, hive, kafka, spark etc

Given that you already have 10+ years experience in System admin in software domain, there are some things to really keep in mind:

  1. Opportunity cost: Whether making the transition and time taken in learning these skills makes sense rather than just continuing with your current role/ progressing in it
  2. Interest level: Are you interested in prediction based problems like Stock market predictions, sentiment analytics, recommendation system etc
  3. Coding: You NEED to code in this profile, there is no way out
  4. Business Acumen: To be a good data scientist you need to have VERY GOOD business acumen of what works and what doesn’t? Simply thinking I will mine everything from data doesn’t work in practice

For starters you can enroll to some of the courses on courses.analyticsvidhya.com to have a look.