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

谢谢K大大 受教了

谢谢分享！现在就看自己努力啦

看起来很牛的样子，不管怎么样先mark

我只能说 K神我真的是爱你 你相信我

Project proposal

1. it's about selling a vision, a huge, beautiful vision
2. you will NOT get there in one leap
instead define MVP, then gradually refine
define the steps that will get you there
3. talk about selling points, benefits ---
use "business impact", always, if possible
what was mentioned the most in the last few high level meetings?
yeah, that, exactly that, is what your proposal will address (i am sure you can find a way for the proposal to address it at least in SOME way, otherwise why do you wanna work on this anyway)
4. do some estimate of cost, ask for resources

dont' group send it out, before you privately show it to a few that are close to you to get feedback

.--very carefully consider feedback, and address them

when you really send it out, it should be quite ready for sunlight.

always sit on it for a couple nights - you will find better phrases/better ideas/selling points

When possible, mock up something visual. No need to go high fidelity here -
mock the FE with wireframe drawings,
or mock with ppt/keynote
or even pencil on paper, and mock interaction with a gif, or shoot a video of switching out paper mocks


====
why would a DS do this?
if you wanna be in the driver's seat, instead of reacting to people's requests, you gotta learn to do this sooner or later
if you want to work on a long term project
if you want to drive for change
if you want resource
if you want a promotion
if you want more people on your team (so you get another promotion)
you gotta learn to do this --- and it's easier than deriving a sequential test or analyzing time series with gnarly trends and outliers.

太棒了 谢谢分享

多谢K姐，下面写一下我未来半年（4-11月）的计划

Background： 生统4th year Ph.D:
目前情况： R: expert (developed 3 packages); SAS: intern experience with data management & GLM / GLMM / GEE, SQL，Shell script, familiar with MATLAB，Hadoop
需要学习/加强：
1) Python(学习了一个月，希望能达到可以用Python熟练处理和分析数据)
2) C++(have basic knowledge，希望可以用C++/ Python 过leetcode题目)
3) Algorithm(上过一门算法课，不过忘得差不多了，现在准备复习一下开始刷leetCode，重点在easy)
4) 熟练MapReduce by Hadoop and Spark. （学习Scala）

非常有用啊~~~谢谢谢谢~

天了噜 太厉害了

spark 是好东西，越用越喜欢， 大家都应该学习下。