Censored Estimation (Capacity Utilization)

[fmramos]

Hi all!

I'm used to estimate some capacity utilization level in Brazil. Of course, the level is between 0 and 1 (0 to 100%). I've been using Truncated Data Estimation.

I'm reasonable smart with techniques, but definitely I'm not academic. I'd like some help for choosing between Normal, Logistic or 'Extreme Value' distributions of errors.

Anyone has some experience in this field?

The choosing of one of distributions, I've tested, generates important differences in coefficients and, of course, forecasts.

Tks in advance,

Fabio

[trubador]

I think the best approach would be to let the data speak for itself. You can carry out a similar exercise to the one discussed here.

[fmramos]

Really Tks, Trubador! I'll try some of the tips u gave in that post; using some LR test. If I find something interesting, I'll send to all. TKs.