Data Scientist 炼成记录-更新完毕2018年12月 | 机器学习练成记录 - 已开新帖

Mark...大赞一亩三分地

大牛啊！

很有帮助的帖子，谢谢

回复马克一下

mark----------------------------------------

Some update:
spent the last couple of months focusing on python (general) programming.
scipy/numpy/pandas and general programming are sufficiently different -- knowing the former won't help you on coding interviews at all
个人用R做分析比用py多一点，不过也没有强烈preference, both have nice tools/vis, R is still better for "deep sh*t" in stats, Py merges with other parts of the system easily.

Finally understand enough about web dev to work on backend DS or hacking projects  - hands on project is the only way to learn.
also spent some time in code structure, tests, intermediate topics

still can't 刷题 in leetcode in any decent manner -- coding面一直强烈短板，依靠其他方面的deep expertise掩盖过关。
java/scala skills are still dismal，至今没用上过

lost some interest in research type, deeper stats
became more interested in solving real life problems with data and code, the data part may or may not involve deep stats
strategic thinking - what is the best approach in terms of cost and benefit and maintainability? 80-20 rules: MVP and iterate - which method in my toolset can solve the problem with the best tradeoff of effort and reward? when is a good time to go the complex route?

I think it's a mentality change - solving real life problems with technology, vs playing with deep/cool stuff in tech that happens to solve some problems as a side effect

Also got more interested in system design as a result -- sexy machine learning algo/model selection and validation plays a very small role in practical systems -- not that it's not important to get things right in principle (it still has to be right), but having the right system: extract and store features (feature engineering), efficiently use memory, predict/classify in a timely manner, etc are probably 95-98% of the work, except for at very large companies with a large team of PhDs, which celebrate 0.1-1% improvements.

mark~~~~

新人报到一下

我现在是在加拿大读Bioresource Engineering的，主要做modeling，但以后想找data analysist类型的工作，然后已经开始自学Data science了。现在主要用Python来刷leetcode的题，还停留在easy水平。SQL自学完了，现在在coursera上不断刷machine learning和algorithm的课程，有点迷茫，各位能指点一下应该怎么学么？感觉小白一头扎进来到处都摸不清。而且因为我专业跟statistic或者CS都没什么关系，找Data方向工作的时候会成为劣势和阻碍么？谢谢各位帮忙啦！

ccx-fdu 发表于 2016-1-12 11:44
.. 我现在是在加拿大读Bioresource Engineering的，主要做modeling，但以后想找data analysist类型的工作，然 ...

So glad to have your kind respond! I will try to do better. THX~