rt6972177/106-2

Jupyter Notebook

陳祈廷的個人Github

Week_1 task:

Show that I can use ipython notebook
Try to use pandas
Viewing datas from 91APP

Week_2 task:

Show that I can make a web crawler
Make a web crawler to get Youtube video information of HowFun
Show the result by dataframe

Week_3 task:

Show that I can use plotly to Visualization
Viewing the data from 91APP
Compare the buying quantity from different payment

Week_4 task:

Show that I can make a wordcloud
Use jieba to cut word
Get the article data of DoctorKoWJ from Facebook API
Make a wordcloud from some article

Week_5 task:

Show that I can conduct TF-IDF to articles
Use jieba to cut word from articles
Separate articles to some groups and conduct TF-IDF
Use wordcloud to see the result

Project_1:

Try to use all tool we have been taught.
Make a web crawler to get web news from CHT & LTN.
Use BeautifulSuop4 to prase the html and get each article.
Use jieba to cut word from each articles.
Use TF-IDF to remove not important word.
Use word to show the Result.
Compare the politics hot news of these two Website.

Proposal_1:

顧客分群以及實驗設計之PPT

Week_7 task:

使用SVM辨識手寫數字
利用sklearn內建的dataset練習使用SVM
使用Isomap進行資料降維，再丟進SVM進行分類
視覺化呈現分類結果

Project_2:

資料清洗之後透過sklearn預測生存與否
利用group函數觀察不同族群間的生存率差異，以找出重要參數(如搭不同逃生艇、不同社會地位者)
嘗試對重要參數使用PCA，但效果不好
訓練出來的模型針對測試資料的預測準確度達97%

Project_2 Answer of prediction csv file

Project_3:

觀察數學&科學成績和投入因素的相關性
在不同欄位的資料計算得的相關係數中，只有在性別的地方出現負相關，值得討論
「數學較好的人科學也會比較好」(相關係數較高)
Try to use plotly and matplotlib to plot the regerssion

Project_4:

從91APP的資料中的訂單，套用Apriori來觀察
利用觀察結果得出針織衫和毛衣圍巾等有較高的機率一起購買

Project_5:

利用上上次作業的影響數學表現的研究資料來嘗試使用PyTorch
資料清洗，包括將有線性關係的文字資料轉換為數字資料，和沒有線性關係的類別型資料的one-hot處理
參考助教的範例架設Model
嘗試對Model做出修改以符合需求

Final_Proposal: