Timeline for How to interpret a confusingly large discrepancy between True Negative Rate and the number of Overspecified Models Selected by LASSO
Current License: CC BY-SA 4.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 30, 2023 at 18:23 | vote | accept | Marlen | ||
| Oct 30, 2023 at 17:09 | answer | added | EdM | timeline score: 1 | |
| Oct 27, 2023 at 10:49 | history | edited | Marlen | CC BY-SA 4.0 |
edited body
|
| Oct 27, 2023 at 6:28 | comment | added | Stephan Kolassa | I absolutely understand your position, one more data point on why it makes sense to get the statistician on board before collecting the data (and actually, when designing the study. Unfortunately, I don't have any input on your actual question, I am sorry. Good luck! | |
| Oct 27, 2023 at 2:22 | comment | added | Marlen | I agree with you there Stephen, but again, I am just a lowly 2nd author on the 2nd draft of a prior working paper which the lead author wrote by himself. So I don't make the rules around here! Furthermore, I started on this project in July of 2022, and this is the very last step for me right here. | |
| Oct 24, 2023 at 8:52 | comment | added | Stephan Kolassa | TPR/FPR and friends are very problematic for all the same reasons as accuracy. Also, they depend heavily on what threshold you are using. Since you do not address this point at all, I suspect you are using a threshold of 0.5, and I would very much encourage you to think about whether this is really what you want. | |
| Oct 24, 2023 at 8:43 | history | asked | Marlen | CC BY-SA 4.0 |