data science 跟 software engineer的比较

An engineer (or your manager) will think:
you only crunch numbers, and didn't do sh*t
you didn't launch product/ship production code/ship quality production code
therefore you didn't do sh*t, and don't deserve to be paid higher...

seriously, people pretty much told me this -- in my face ... and no, not from people that personally hate me, just engineers who share their honest opinion....
what people don't see, they don't value - so either make slides (which are paid even less) or ship code

if you can ship production code and do damn good statistics, then you are golden

shiloh00 发表于 2016-3-3 16:43
感觉面试题目就不一样 我面了个data scientist 直接蒙蔽了 上来就问我什么事overfitting 如何避免overfitti ...

是的 应该是很简单的了 可惜我不用r TT 但是我知道r很简单 所以希望面试官可以给我放点水

overfitting has nothing to do with R....

小K 发表于 2016-3-4 03:27
overfitting has nothing to do with R....

0 0 我没有说overfitting和r有关系好么 我是说R很简单 overfitting是ml的知识

shiloh00 发表于 2016-3-4 03:36
0 0 我没有说overfitting和r有关系好么 我是说R很简单 overfitting是ml的知识

问了 我说我不会 我就用C＋＋ 她说那好的 R很容易学 只要有编程背景就好 呵呵哒

shiloh00 发表于 2016-3-3 12:02
问了 我说我不会 我就用C＋＋ 她说那好的 R很容易学 只要有编程背景就好 呵呵哒

but she still failed you for R syntax?

小K 发表于 2016-3-4 04:28
but she still failed you for R syntax?

Nope! lol

Skill set区别挺大的，统计出生的做DS很自然，CS出生的还是SDE优先吧 。

就个人经验来说，DS做到senior以后非常需要人际关系、口才和reputation，和SDE的life style非常不一样，比如我自己现在30%左右的时间要东西海岸甚至欧洲的飞去参加client meeting，组里一个中国人都没有，一个月也发不了几个diff……

"=.=||| 关键是你得知道啥时候用啥"

-- re
比如这贴首楼其实是个simpson's paradox问题。。。。

看平均值不能了解整个distribution