Profile_bird

Hey there! Snosrap is using Twitter.

Twitter is a free service that lets you keep in touch with people through the exchange of quick, frequent answers to one simple question: What's happening? Join today to start receiving Snosrap's tweets.

Already using Twitter
from your phone? Click here.

Snosrap

  1. TAA-USA Typhoon Marokot donation information http://www.taa-usa.org/ 9:39 AM Aug 11th from web
  2. Got some sense about different longitudinal data analysis techniques 8:14 AM Jun 10th from web
  3. The idea of Mixed models is intuitive, but it involves higher level mathematics. Good thing is that we can leave those to computer. 11:15 AM Jun 9th from web
  4. @NICO1106 你最近很殺喔 8:27 AM May 27th from web in reply to NICO1106
  5. @NICO1106 You can think 'Mean' as n. or adj. Either way is fit.... 5:52 AM May 15th from web in reply to NICO1106
  6. Luckily, I am not one of the "we" yet. 10:02 AM May 14th from web
  7. We, statisticians, are mean lovers. 8:48 AM May 14th from web
  8. The goal of this study project is going to explore the answers of these three difficulties in literature. 3:56 PM May 10th from web
  9. The last difficulty is that after matching, how do we evaluate the validity of the matching? What kind of sensitivity analysis we should use 3:55 PM May 10th from web
  10. Another difficulty is that after getting the propensity score, which matching method we should use to find similar persons b/w conditions. 3:54 PM May 10th from web
  11. However, the difficulty lays in how many and which covariates we should include when calculating the propensity score. 3:52 PM May 10th from web
  12. The beauty of propensity score is that it summarizes the set of covariates with a single score for matching. 3:51 PM May 10th from web
  13. The idea of counterfactual is to find similar persons in different conditions based on a set of covariates. 3:50 PM May 10th from web
  14. 圖片的來由：台灣建國也許就像凱茲一樣受到祭品的詛咒卻永不認輸的提著大刀為自己的命運而奮鬥。 2:20 PM May 10th from web
  15. Starting a new research topic: Counterfactual 1:59 PM May 10th from web
  16. But the design is weird enough to fit a survival model, since at each time point the disease status change back to non-disease. 6:48 AM May 8th from web
  17. The GEE model works for repeated measure of binary or categorical data. However, with an epidemic design, I am not sure if GEE model work. 6:47 AM May 8th from web
  18. 迴歸分析犯了錯誤，得全部重跑。還好，只需要花十多分鐘而已 8:11 PM May 4th from web
  19. 今天繼續把剩下50幾個模型跑完。又花了4個小時。結果跑另外一組分析，也是上百個模型，只花了10分鐘。迴歸分析還是比較簡單啊。明天得開始整理跑出來的東西了。 4:19 PM May 3rd from web
  20. 週末在家，用遠端遙控辦公室的電腦狂跑上百個模型。可以想像辦公室電腦在哀鳴。可是已經跑了六個小時了，還有100個模型要跑。 5:31 PM May 2nd from web