« first day (3307 days earlier)      last day (384 days later) » 

2:37 AM
Well, it didn't last long.
8 hours later…
10:07 AM
Hi guys, who of you has worked with dependent samples?
I posted also a question about but it didnt generate a lot of attention
I am wondering about the datasplitting procedure of the R-Caret Package groupKFold. Can I remove the ID column after training. I thought previously that it should be no problem after the folds are created. However a comment of friend of mine made me insecure about this
Q: Data splitting with caret: Can we remove the ID Coloumn after folds are created?

PythonBeginnerso we have a dependent sample (two observations for each participant). To prevent data from one participant being in the training set and in the unseen fold in cross-validation we used the groupKFold method from the caret R-Package. I was wondering if it is correct that I remove the ID column aft...

Are female hurricanes deadlier than male hurricanes?

The Dutch meteorological institute is gonna start naming storms. This somehow brought the attention to this research on difference between female and male hurricanes. Being skeptical I had to check it out. Was there some false representation of statistics in the research? Well check it out. What do you people think about the use of the numbers of degree of freedom when only five female/male names are being tested?

"As expected, hurricanes with male names (Arthur = 4.246, Cristobal = 4.455, Omar = 4.569, Kyle = 4.277, and Marco = 4.380)
10:30 AM
Furthermore: The 4 most deadliest hurricanes of the past decades from their data-set (kudos for sharing it) were female (Diane 1955, 200 deaths; Camille 1969, 256 deaths; Agnes 1972, 117 deaths; Sandy 2012, 159 deaths). However, in the early days all hurricanes had female names and there can not have been this variation in perception between male and female names.
2 hours later…
12:52 PM
@MartijnWeterings That paper has been heavily criticized for poor analysis. I wrote about that a bit and linked to the critiques in this answer: stats.stackexchange.com/a/418019/121522
2 hours later…
2:46 PM
@mkt thank you for pointing out some links. Or actually not. The discussion sketches a horrible situation. The criticism avoiding response like publish.illinois.edu/shavitt/files/2013/07/PNAS-Reply.pdf cuts away a big chuck out of my last, and already small, remaining idealistic hopes for science and society.

« first day (3307 days earlier)      last day (384 days later) »