原文: 21 Reason why you should NOT become a Data Scientist

數據科學家

這禮拜看到一篇以圖帶文的21 Reason why you should NOT become a Data Scientist，他提了一些諷刺的觀點來反映現有的數據科學盲點，是蠻值得我們在使用這些模型及工具時應當注意且回頭來檢視自己的研究是否應當修正或納入的觀點。我是相當肯定機器學習或大數據的分析及理論方向，所以我將 21 點濃縮並在研究及學習中持續檢視。

** 6 觀點 **

1. 數據科學應當是提出相對的觀點建議而非絕對，因為我們是按現有的資料來作預測，就如同統計分析應當註明正確及錯誤率。
2. 不要操作或改變資料來影響微小的結果差異。
3. 儘可能找到實質影響分析結果的原因，而不是只有類神經的黑盒子。
4. 能夠說明各演算法的優劣
5. 向領域專家學習，而非用數據教導他們
6. 數據科學家持續學習是必要的，但玩樂也是重要的

21點為什麼你不該成為數據科學家

1. Trump: When every data scientist in the world got it wrong!
2. At times you’ll get it horribly wrong. Did I just say ‘wrong’? Epic FAILURES!
3. Ever participated in Kaggle? This is how I feel when you have to spend days getting that small increase in the 5th decimal place.
4. And in case, I got awesome results from a black box model – this is how it feels!
5. Torture the data and it’ll confess what you want it to.
6. Why study so many algorithms when XGBoost always does the trick for you!
7. Automation! My job is to make machines replace me.
8. I must learn the languages that are going to pass out in 10 5 years anyway.
9. Looks like I’m the only one who calculates the p-value of getting an increment everyday.
10. I get bashed by the CEO daily while everybody stands and watches.
11. Nothing is impossible until you start to explain Data Science at a social gathering.
12. Human thinking ends at 3-D. My work starts at 100-D.
13. Carrying cool laptops is a dream. Carrying servers is a necessity.
14. Only god knows the future. Whom am I to predict.
15. Astrologers have been doing it for years.
16. Why spoil weekends over a hackathon/competition? Coldplay?
17. I am expected to teach domain expertise to domain experts.
18. Any my expertise depends on where I’m giving the interview.
19. Don’t doubt me. Alternative hypothesis stands true!
20. No one knows who a Data Scientist is?
21. The world is a strange place. And believe me, it’s not at all like what you think.