Data Science之你問我地答

香港玩CV最好係做medical果邊

你入左行基本上就係香港無乜對手

咩叫識build model

CNN from scratch定係model.fit()

用scikit learn 個train test split嘛

model.fit()夠做啦
反正係preferably

train = data[:70]
test = data[70:]

點樣入行？
自學緊

好撚認真做係應該做一次vector encoding後分一分cluster睇下distribution既

你機多就分幾個fold去做, 無機無時間就skl split算

skl有用train test split有得揀state 同揀random shuffle

睇cluster目的係make sure唔同features既distribution係train test sets入面分別差唔多？

係, 呢樣野可以分完做一做gaussian mixture model

不過我從來無做過

讀master, 或姐做data類IT工再轉過去
我覺得後者會容易啲，始終entry DS role真係好少

睇咩tree, CART/C4.5 ,計Gini/entropy
不過依家啲lib咁醒幾句statement都做到哂
反而tune堆parameter真係唔易

same
依啲就係第一頁我話我覺得自己似風水佬既原因

不過理論上random split完個distribution比例應該會平均，不過要夠多data同埋唔好咁黑仔

data類it工？

好多 data scientist 番工連data 都冇又要夾硬諗D 有revenue 既野出黎, 冇野搵野黎做

自學有咩text book推介
要裝咩software
有無website介紹/notes share

姐係data analyst, data engineer, BI report個啲，最好同data scientist work既posts

我賭波呀, 不過呢排無得玩

text book唔知，推介睇啲online course, 好多唔洗錢
software裝anaconda, 用jupyter notebook

“I Dropped Out of School to Create My Own Data Science Master’s — Here’s My Curriculum” by David Venturi https://link.medium.com/vWhBL0pit6

講真有幾多人知每一層layer input output size幾大
點解要用呢個layer

最近睇緊呢本



我上手學TF 係睇stanford TF 教材，想好撚數學就Ian Goodfellow本Deep Learning

純DS唔介意殘體字就西瓜書

實際做野就煲github同skl嘅doc