Comments (14)
Yep, that's a bug, fixed in release 0.7.2. See https://github.com/melff/mclogit/releases/tag/0.7.2
from mclogit.
from mclogit.
Hi,
4 days ago I put up the 0.8 release. The algorithm should be more stable now. Let me know if you still encounter convergence problems. If all else fails, 0.8.2 supports estimation of an overdispersion parameter, which allows for a (rather crude) alternative of dealing with clustered data.
from mclogit.
from mclogit.
This should be fixed by release 0.8.6.1
from mclogit.
I think I identified the bug, which is now fixed in the current release: 0.8.6.3.
from mclogit.
Hello Prof. Elff,
Just a follow up. I estimated again the model and now this problem doesn't occur anymore, but I receive the following messages while estimating a complete model. This is probably related to my data set and you can ignore, but I thought that maybe would be important to send a feedback.
>mblogit(formula = dropout ~ sex + age_group + secondary_school + fam_income + color_race +
+ receive_grant + extracurricular,
+random = ~ 1 | institution/program_id,
+data = stud.db)
Iteration 1 - Deviance = 157998.1
Iteration 2 - Deviance = 147780.8
Iteration 3 - Deviance = 146485.8
Iteration 4 - Deviance = 146400.7
Iteration 5 - Deviance = 146399.1
Iteration 6 - Deviance = 146399.1
Iteration 7 - Deviance = 146399.1
converged
Iteration 1 - deviance = 150283.3 - criterion = 0.1145848
Stepsize halved - new deviance = Inf
Stepsize halved - new deviance = Inf
Stepsize halved - new deviance = Inf
Stepsize halved - new deviance = Inf
Stepsize halved - new deviance = 193007
Stepsize halved - new deviance = 183789.1
Stepsize halved - new deviance = 179251.9
Stepsize halved - new deviance = 177098.2
Stepsize halved - new deviance = 176032
Stepsize halved - new deviance = 175514.9
Stepsize halved - new deviance = 175640.2
Stepsize halved - new deviance = 175837
Stepsize halved - new deviance = 175936.8
Stepsize halved - new deviance = 175986.8
Stepsize halved - new deviance = 176011.7
Stepsize halved - new deviance = 176024.2
Stepsize halved - new deviance = 176030.5
Stepsize halved - new deviance = 176033.6
Stepsize halved - new deviance = 176035.2
Stepsize halved - new deviance = 176035.9
Iteration 2 - deviance = 176035.9 - criterion = 2.816581e-09
converged
Warning messages:
1: step size truncated due to possible divergence
2: Algorithm stopped due to false convergence
Again, thanks a lot for all the help.
from mclogit.
Hello Martin,
Thanks for let me know about the release.
Tried again and now it converged. Just to report some results, when I run the model with limited variables, it produces NaN in the grouping level 2.
>mblogit(formula = dropout ~ sex + age_group,
+random = ~ 1 | institution/program_id,
+data = stud.db)
Iteration 1 - deviance = 166003 - criterion = 0.8051472
Iteration 2 - deviance = 145565 - criterion = 0.05297786
Iteration 3 - deviance = 139727.4 - criterion = 0.02102629
Iteration 4 - deviance = 137573.3 - criterion = 0.008212973
Iteration 5 - deviance = 136641.2 - criterion = 0.002999969
Iteration 6 - deviance = 136343.1 - criterion = 0.000300909
Iteration 7 - deviance = 136278.7 - criterion = 1.705491e-05
Iteration 8 - deviance = 136251.1 - criterion = 5.813921e-07
Iteration 9 - deviance = 136246 - criterion = 1.285311e-08
Iteration 10 - deviance = 136245.3 - criterion = 3.098197e-10
converged
(Co-)Variances:
Grouping level: 1
Estimate Std.Err.
Dropout~1 1.494 4.758
Stopout~1 2.143 3.948 8.221 14.188
Grouping level: 2
Estimate Std.Err.
Dropout~1 2.206 NaN
Stopout~1 1.447 3.139 NaN NaN
Null Deviance: 482500
Residual Deviance: 136200
Number of Fisher Scoring iterations: 10
Number of observations: 219597
Warning message:
In sqrt(diag(vcov.phi)) : NaNs produced
And with the complete model, it doesn't produce NaNs.
>mblogit(formula = dropout ~ sex + age_group + secondary_school + fam_income + color_race +
+ receive_grant + extracurricular,
+random = ~ 1 | institution/program_id,
+data = stud.db)
Iteration 1 - deviance = 165108.2 - criterion = 0.8049309
Iteration 2 - deviance = 143677.5 - criterion = 0.05636877
Iteration 3 - deviance = 137233.2 - criterion = 0.0243093
Iteration 4 - deviance = 134822 - criterion = 0.01024263
Iteration 5 - deviance = 133809.7 - criterion = 0.003024441
Iteration 6 - deviance = 133378.3 - criterion = 0.0007326398
Iteration 7 - deviance = 133226.5 - criterion = 7.890494e-05
Iteration 8 - deviance = 133181.8 - criterion = 4.898939e-06
Iteration 9 - deviance = 133171.1 - criterion = 3.000677e-07
Iteration 10 - deviance = 133168.8 - criterion = 2.003485e-08
Iteration 11 - deviance = 133168.3 - criterion = 1.462663e-09
converged
(Co-)Variances:
Grouping level: 1
Estimate Std.Err.
Stopout~1 1.638 3.661
Dropout~1 2.356 4.344 6.334 10.948
Grouping level: 2
Estimate Std.Err.
Stopout~1 2.871 0.2844
Dropout~1 2.288 5.060 0.2637 0.3159
Null Deviance: 482500
Residual Deviance: 133200
Number of Fisher Scoring iterations: 11
Number of observations: 219597
A very newby question: how the estimation is determined by the package and it is possible to change these parameters?
Again, thank you very much.
from mclogit.
Understood! Thanks!
from mclogit.
Hello Martin,
In the newest version of the package, I fitted the first model that I posted here, without any modifications, and I'm getting this message:
Error in res[[i, j]] <- t(x[[i, j]]) : [[ ]] subscript out of bounds
I modified the method and estimator arguments to other options and used more or less variables, but I receive the same message.
from mclogit.
Hi Martin, thanks for the return!
Tested again and now I get this message:
Error in x[[j, i]] : subscript out of bounds
Followed the same tests as before. This happens with a model with two random effects. When I exclude one of them, It works.
from mclogit.
Dang, I thought I got it. Can you please show what traceback()
shows if immediately called after the error message has been shown? That would help me in bug-hunting.
from mclogit.
Sure! Like this?
Error in x[[j, i]] : subscript out of bounds
> traceback()
8: t(x[[j, i]])
7: bMatTrns(ZWX)
6: cbind(blockMatrix(XWX), bMatTrns(ZWX))
5: rbind(cbind(blockMatrix(XWX), bMatTrns(ZWX)), cbind(ZWX, ZWZiSigma))
4: structure(rbind(cbind(blockMatrix(XWX), bMatTrns(ZWX)), cbind(ZWX,
ZWZiSigma)), class = "blockMatrix")
3: PQLMQL_innerFit(y.star, X, Z, W, d, groups, offset, method, estimator,
control)
2: mmclogit.fitPQLMQL(y = Y, s = s, w = weights, X = XD, Z = ZD,
groups = groups, method = method, estimator = estimator,
control = control, offset = offset)
1: mblogit(formula = ind_EVASAO_TER ~ dep_SEXO + dep_IDADE_FAIXA +
dep_ORIGEM_ESCOLAR, random = ~1 | COD_IES_2016/COD_IES_CURSO_2016,
data = bd.subset)
from mclogit.
Just tested again and it's working! Thanks a lot.
from mclogit.
Related Issues (20)
- Error with "predict" command in newest versions HOT 11
- `*tmp*`[[k]] : subscript out of bounds HOT 2
- Incorrect predictions from mblogit when predictors are scaled HOT 3
- mblogit model comparison with anova does not calculate degrees of freedom from random parameters HOT 3
- Problems with model definition HOT 1
- Error when using nested random terms HOT 1
- Error `'a' (<dimension1> x 1) must be square` HOT 3
- Error from `predict.mmblogit()` HOT 2
- Incorrect result by `predict.mmblogit()` in case of multiple `random` formulas HOT 3
- Syntax for increasing iterations of mblogit model HOT 1
- systematic error with nested model HOT 2
- Frequent error in mblogit "Error in solve.default(Information) : system is computationally singular: reciprocal condition number" with models that run OK in nnet::multinom HOT 4
- Error "Error: no valid set of coefficients has been found: please supply starting values" for model that runs OK in nnet::multinom HOT 2
- [Question] Interpreting and plotting a mblogit model
- Feature request: allow some form of regularisation in multinomial fits? HOT 2
- Problem with random effect: In sqrt(diag(vcov.phi)) : NaNs produced HOT 2
- Error in formula HOT 5
- `summary` printing odd behaviour depending on response level names HOT 1
- Equivalent functions from lme4. HOT 1
- `Warning: Inner iterations did not coverge` and an advise about model structure HOT 6
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from mclogit.