In reality, I think that Big Data contains three dimensions:
- Computational science: Getting the data, cleaning the data and writing code for the stats. This is a combination of database knowledge (SQL & others), scripting (python & others) and statistical programming (R & others).
- Statistics: What does the data mean? What can you infer? What can't you infer?
- Domain knowledge: What are the needs of a consumer products company? Or telecom company? Or financial services company.
A "Big Data" person needs all three. Most computational science programs focus mostly on the first, only a little on the second, and not at all on the third. I'd argue that academic programs should go heavier on the stats, as that's the field that will change the least, and will be the hardest to learn later. You can always learn a scripting language later, and much of the domain knowledge can be picked up outside of school.
This is a good breakdown. I was hoping there would be a difference between the two other than just a reason to make a program for a buzz word. Thanks for this.
6 comments
[ 0.20 ms ] story [ 23.9 ms ] threadIn reality, I think that Big Data contains three dimensions: - Computational science: Getting the data, cleaning the data and writing code for the stats. This is a combination of database knowledge (SQL & others), scripting (python & others) and statistical programming (R & others). - Statistics: What does the data mean? What can you infer? What can't you infer? - Domain knowledge: What are the needs of a consumer products company? Or telecom company? Or financial services company.
A "Big Data" person needs all three. Most computational science programs focus mostly on the first, only a little on the second, and not at all on the third. I'd argue that academic programs should go heavier on the stats, as that's the field that will change the least, and will be the hardest to learn later. You can always learn a scripting language later, and much of the domain knowledge can be picked up outside of school.