chevrotin

Thanks for your quick reply. It isn't small. The dataset has over 1000 individuals, 12 choice sets if I recall correctly. But it is not unusually big for what I...

Thank you both for your wonderful contributions. On Fri, Dec 15, 2023 at 5:18 PM Cristian Arteaga ***@***.***> wrote: > Hi everybody, I just wanted to chip in to mention...

Hey John, I'm diving into how WTP Space is implemented on xlogit now. It would be awesome if different starting values can be tried as in logitr now. Would it...

Thanks again. Combining xlogit and logitr would get us lightning speed and useful features. It would be very exciting for choice modelers. Keep up the good work. On Mon, Dec...

Thanks Cristian. The code works. I managed to estimate a WTP-space model with my data. Providing starting values, derived from a preference space ML model, seems to do the trick....

Ah.. thanks. Glad that you have already solved that. Wasn't reading careful enough. Will try the batch function.

If you need a helping hand for documentation, I'll be glad to help.

Alright.. Will do

https://colab.research.google.com/drive/1bjuh4-oAcsz5m9rN4uYnmQtqYPeC-Vwf?usp=sharing @arteagac Please see the examples of both starting values and using batch_size to estimate with ndraws = 50000. Note that the results differ a bit to LOGITR. I think...

Glad to hear from you, John. Here are some of my thoughts: @jhelvy It's true that GMNL is not necessary for WTP-space. The plain ML can do it. In my...