I really enjoyed the discussion. Yes, statistics is difficult. But it needs to be understood to be used correctly. I'd like to hear what this group says now, currently about the combination of programming and statistics. Data science has been included as a degree at my university, however it is still so difficult to understand how the data scientist actually is defined. However, after 3 years of statistics, when you get to the end of all the theoretical learning and finally get into putting some stuff in R, it is really amazing to see how everything comes together and how modelling can finally be done. It is something truly satisfying. And I think the rigorous statistical background is quite necessary to get there and truly enjoy and understand the experience. I think the true data scientist is someone who enjoys the process and beauty of statistics, as well as the processes and beauty of computer science. Not someone who leans to either side, but rather enjoys putting them together.
Absolutely brilliant stuff, with some brilliant minds. Thank you. 5 years on and the debate still stands. It would be great if the same panel got together again in 2021 for a new debate. :)
It is interesting that a data scientist should have a background in both statistics and computer science. Both these fields are rigorous. Therefore, I support an idea that division not only of labour but also knowledge is the most effcient way for the data science field how to evolve. Actually these two should not be brought together, but left seperated.
It's true that a data scientist is a better statistician than a computer engineer and better computer engineer than a statistician, BUT, it's worse computer engineer than a computer engineer, and worse statistician than a statistician.
I am a statistician and I am a much better programmer than most data scientists. No idea where this old view comes from where people think statisticians can't program. What I also find strange is when some job includes some programming as a tool, it is immediately associated with computer science. While I was modeling astrophysical processes, I was a computational physicist, NOT a software engineer.
From the floor questions, its sounds like the statistics teaching community could take a leaf from the way graduate business schools teach using the case method. Perhaps this might be a new line for HBS?
@@denzokyedravk see.. If you want to learn to be a professor or just gain knowledge statistics is fantastic.. But irl if you want a job i would suggest data science
"Big Data" means data volumes too large to handle conveniently with current technology. It's been around for a very long time; it's just the orders of magnitude that change. The question is whether there really is a separate trade of "Data Scientist"? Maybe there are already people who can do what is wanted, just using different names for the job?
Instead of placing all these disciplines into one and calling it data science just create a team of statisticians, analyst, computer scientist and any other field to work on the project at hand
Sure, I think in large scale projects that is still the case. Yet the need for a generalist, knowing of all those fields to a solid degree might still be in need to be on the upper end of planning, and managing the processes.
Great panel, but dont think they have ever delivered a solution in production working with crazy, demanding and non-mathematical Banking or any other client. They are great in knowing the theories but real world is totally different than theories. To be a good data scientist, you need only 2 things - 1. "I can do it" attitude and 2. Common sense. Nothing else.
Looks like you mainly do recommendation system. Statistics comes in translating a business problem into analytical problem, linear algebra helps in solving all the big matrix calculations.
This viewpoint of them versus us does not feel productive. Linear algebra itself is enjoying the fruits of statistical methods via randomized numerical linear algebra algorithms to speed up fundamental operations like matrix multiplication and solving linear systems. Check out e.g. recent work on oblivious subspace embeddings. So, my point is, classical linear algebra has been fundamental in supporting implementations of statistical methods, and modern linear algebra benefits from advanced statistical analyses.
I really enjoyed the discussion. Yes, statistics is difficult. But it needs to be understood to be used correctly. I'd like to hear what this group says now, currently about the combination of programming and statistics. Data science has been included as a degree at my university, however it is still so difficult to understand how the data scientist actually is defined. However, after 3 years of statistics, when you get to the end of all the theoretical learning and finally get into putting some stuff in R, it is really amazing to see how everything comes together and how modelling can finally be done. It is something truly satisfying. And I think the rigorous statistical background is quite necessary to get there and truly enjoy and understand the experience. I think the true data scientist is someone who enjoys the process and beauty of statistics, as well as the processes and beauty of computer science. Not someone who leans to either side, but rather enjoys putting them together.
We need these series on Netflix right inmediately
Absolutely brilliant stuff, with some brilliant minds. Thank you. 5 years on and the debate still stands. It would be great if the same panel got together again in 2021 for a new debate. :)
Intellectual power exhibited by Hand is on another level .
One of the most interesting discussions about this topic. Insightful!
23:00 - 24:50 Such an accurate and interesting observation....and so perfectly said.
I enjoyed the comparison between Statistics and Mathematics by senior staff members both from academia and industry.
Tushar Kale I agree. It was an cool to hear the correlation
It is interesting that a data scientist should have a background in both statistics and computer science. Both these fields are rigorous. Therefore, I support an idea that division not only of labour but also knowledge is the most effcient way for the data science field how to evolve. Actually these two should not be brought together, but left seperated.
Giving this a thumbs up for the statistician!! 👏🏻👍🏻
stupid isreal 👍🏻
I'll prefer being dead than having to listen to this guys for 1 and a half hour
I don't know how I have survived
I thoroughly enjoyed this talk, and great learning too. Thank you RSS 😄
It's true that a data scientist is a better statistician than a computer engineer and better computer engineer than a statistician, BUT, it's worse computer engineer than a computer engineer, and worse statistician than a statistician.
I am a statistician and I am a much better programmer than most data scientists. No idea where this old view comes from where people think statisticians can't program. What I also find strange is when some job includes some programming as a tool, it is immediately associated with computer science. While I was modeling astrophysical processes, I was a computational physicist, NOT a software engineer.
51:00 Stats is its own discipline or else math funding issues
From the floor questions, its sounds like the statistics teaching community could take a leaf from the way graduate business schools teach using the case method. Perhaps this might be a new line for HBS?
I'm having hard time choosing between msc stats or msc data science after bsc stats
Which was the better option?
Please advise. I am faced with the same conundrum?
@@denzokyedravk okay I've completed MSc statistics and i would suggest you to go for data science
@@pallaviharishchandre3021 Why though? Why would you recommend data science over MSc in Statistics?
@@denzokyedravk see.. If you want to learn to be a professor or just gain knowledge statistics is fantastic.. But irl if you want a job i would suggest data science
All subsets of Maths. :)
23:20 UG stats
9:23 he definetely brome the ice
big data also involved data , anything anywhere you want statistics will involved so big data is the small part of statistics
"Big Data" means data volumes too large to handle conveniently with current technology. It's been around for a very long time; it's just the orders of magnitude that change. The question is whether there really is a separate trade of "Data Scientist"? Maybe there are already people who can do what is wanted, just using different names for the job?
104:32-113:45 An ideal curriculum for a data scientist and staticians..?
since people can do data science knowing just some stats, some math, some programming; data science is a subset of none of these.
Instead of placing all these disciplines into one and calling it data science just create a team of statisticians, analyst, computer scientist and any other field to work on the project at hand
Sure, I think in large scale projects that is still the case. Yet the need for a generalist, knowing of all those fields to a solid degree might still be in need to be on the upper end of planning, and managing the processes.
vengo de la uji :(
The girl don’t seem to understand the scientific part of this conversation
Do you need exceptional hacking skills to be an awesome Data Scientist?
I don't think so
Hackers are specialists in cyber security. Data scientists are good programmers and understand quite a bit of machine learning.
Hacking means to dominate different realms in this context. A data scientist must be the joint of Math/Stats/Machine Learning/Subject Matter
Great panel, but dont think they have ever delivered a solution in production working with crazy, demanding and non-mathematical Banking or any other client. They are great in knowing the theories but real world is totally different than theories. To be a good data scientist, you need only 2 things - 1. "I can do it" attitude and 2. Common sense. Nothing else.
Data Science is more about linear algebra than statistics !
Looks like you mainly do recommendation system. Statistics comes in translating a business problem into analytical problem, linear algebra helps in solving all the big matrix calculations.
Synaps Diallo as a mathematician and data scientist couldn’t disagree more with you
This viewpoint of them versus us does not feel productive. Linear algebra itself is enjoying the fruits of statistical methods via randomized numerical linear algebra algorithms to speed up fundamental operations like matrix multiplication and solving linear systems. Check out e.g. recent work on oblivious subspace embeddings. So, my point is, classical linear algebra has been fundamental in supporting implementations of statistical methods, and modern linear algebra benefits from advanced statistical analyses.
Good
I hate this