How to do grouping of analytics data?

What kind of analytics/data science job is the best?

• What are the pros and cons of analytics companies in India which offer analytics as a service and those offering those as a product? There are 5 broad types of Analytics firms in India 1) KPOs like Accenture, WNS, EXL services etc. in which analytics is a part of their outsourcing services. 2) MuSigma, Fractal Analytics, Opera Solutions, Marketelligent : Niche Analytics companies but still offering it as a service to clients 3) Captives like Bosch, Reliance, GE, Banks which have their in house analytics and data science practices, but analytics is not the core part of their business 4) E-commerce firms like flipkart and amazon,24x7 etc 5) Startups focused on offering data analytics and machine learning as a product or software What are the pros and cons of working in these setups?

• Answer:

  I believe more important than looking at companies is what kind of job is offered to you. Make it clear during interview rounds. Now as you know, Data Science is a vast term that now covers almost every thing related to data processing - whether it has that pattern recognition (science) component or not. 1. KPOs usually hire analysts for top level data crunching. If you are an expert in Machine Learning and Optimization Methods, these are not the right fit for you. Pro: Easy to get in if you are not an expert. Con: You won't learn a lot once you are in. You would stick to the tools you'd be trained in rather than reaching out and finding what all exists in the world. 2. Niche Analytics companies have multiple teams that are assigned to totally different kind of projects. MuSigma definitely has been working on real Data Science stuff that previous category. There are companies which are solving challenging problems creatively using Data Science. Look around. Machine Learning is in almost every field now. If you are able to find a company with 10-50 people, merrily join that. Small companies have their own privileges. You feel a sense of belonging. Pro: Good environment, great people from good universities, lot to learn. Con: Can't generalize. You can't be sure which team you might be tagged with. Maybe you get in as a data scientist but keep writing SQL queries for next three years. 3. Finance Institutions usually have great analysis wings. But they usually hire some other third party companies for advanced predictive analysis jobs. You again forgot those like Bloomsberg, Factset etc. which have great teams working on Machine Learning applications in finance analysis and prediction. However I'd stick to banks here. Pro: You'll get lot of domain knowledge. Con: More of statistical analysis, but away from real applications of machine learning (though exceptions exist). 4. Ecommerce companies which have crossed the mark that separates the new startups and next phenomenon have team with real experts. I'm aware of Data Science teams at Amazon, Flipkart and Housing. They hire top-school graduates with solid expertise in Machine Learning and regular algorithm skills. Work in their Data Science team is quite impressive. They are working on the latest technologies, creating new algorithms just to make us buy more of what we want. Pro: Great teams, you will be working on crunching rapidly coming data with latest tools and implement state of the art algorithms Con: Companies with capital in billions and thousands of employees start building bureaucratic management framework. Again, multiple teams might be there - and you'll have to find the right fit. 5. ML Based startups are on the peak. (Gartner's cycle) Some of them have founders as top level experts from giants like Google, Microsoft, Amazon etc.  That's a very important factor while looking at a company - founders and initial team build a culture that will follow over years and decades. If founders are great, that's a great place. Pro: You rapidly learn a lot. If the team is great, your learning is enhanced. Small group - so rather than colleagues, you become a family. No office games. Everyone is focused on building or improving the product. Con: If you are habituated to the standard 9 to 5 work with given deadlines in teams of five, you will find it hard to survive. The work is relatively more challenging than other categories. I would advise mentioning exact skill-set rather than the vague Data Science term before looking forward for such classification. Usual definition used in top companies and academia involves good amount of mathematical skills, thorough knowledge of state of the art machine learning algorithms, programming, knowledge about tools to crunch huge data and necessary hacking skills to get work done.