I have a concern with a GLMM I am running and I would be very grateful if you could help me out.
I am modelling the factors that cause a frog species to make either a type 1 or type 2 calls. I am using a GLMM logistic regression. The data from this were generated from recordings of individuals in frog choruses of various sizes. For each male in the dataset, I randomly chose 100 of his calls, and then determined if they were type 1 or 2 (type 1 call =0, type 2 call =1). So each frog is represented by the same number of calls (100 calls), and some frogs are represented in several choruses of different sizes (total n= 12400). The response variable is whether each call in the dataset is type 1 or 2, and my fixed effects are: the size of the chorus a frog is calling in (2,3,4,5,6), the body condition of the frog (residuals from an LM of mass on body length), and standardized body length (SVL) (body length and body condition score are not correlated so no VIF issues). I included frog ID and the chorus ID as random intercepts.
Model results
The model looks fine, and the coefficients seem sensible; they are about what I expect. The only thing that worries me is that, when I calculate the 95%CI for the coefficients, the coefficient for body condition has a huge range (-9.7 to 6.3) (see screenshot). This seems crazy. Even when exponentiated, it seems quite crazy (0 to 492). Is this reasonable?
This variable was involved in a significant interaction with chorus size; does this explain a wide CI? Or does this suggest my approach is flawed? Instead of having each male equally represented in the dataset by 100 calls in each chorus he is in, should I instead collapse that down to a proportion (e.g. proportion of type 2 calls out of the 100 randomly selected calls for each male) and model this as a poisson regression or something? Is the way I’m doing my logistic regression a reasonable approach? I have run model checks and everything and they all seem to point to logistic regression being suitable for my data, at least as I have set it up currently.
Thanks for any help you can provide!
Values I get after standardizing condition:
2.5 % 97.5 %
.sig01 2.0948132676 3.1943483
.sig02 0.0000000000 2.0980214
(Intercept) -3.1595281536 -1.2902779
chorus_size 0.8936643930 1.0418465
cond_resid -0.8872467384 0.5746653
svl -0.0865697646 1.2413117
chorus_size:cond_resid -0.0005998784 0.1383067
I am trying to topic extraction with STM.
I have a question on how to determine the optimal number of topics.
kResult <- searchK(out$documents, out$vocab, K=c(7,8,9,10), prevalence=~rating+s(day), data=meta)
kResult$results
plot(kResult)
Numerical values and graphs are output in the output result of the searchK function.
I don't know how to determine the optimal number of topics from this result.
I would like to know how to determine the number of topics.
> kResult$results
K exclus semcoh heldout residual bound lbound em.its
1 7 8.937433 -52.95924 -7.80857 9.328384 -23391733 -23391725 17
2 8 9.090138 -58.20191 -7.793394 8.950438 -23337625 -23337614 20
3 9 9.168978 -61.09091 -7.781923 8.710382 -23296459 -23296447 25
4 10 9.256421 -61.51863 -7.764806 8.504863 -23247891 -23247876 55
kResult plot result
I read the treatise, but couldn't understand what the following values represent.
exclus：Exclusivity of each model.
semcoh：Semantic coherence of each model.
heldout：Heldout likelihood for each model.
residual：Residual for each model.
bound：Bound for each model.
lbound：lbound for each model.
em.its：Total number of EM iterations used in fitting the model.
Also, I don't know what each of the graphs below represents.
kResult plot result
Could someone help me to determine the correct random variable structure in my binomial GLMM in lme4?
I will first try to explain my data as best as I can. I have binomial data of seedlings that were eaten (1) or not eaten (0), together with data of vegetation cover. I try to figure out if there is a relationship between vegetation cover and the probability of a tree being eaten, as the other vegetation is a food source that could attract herbivores to a certain forest patch.
The data is collected in ~90 plots scattered over a National Park for 9 years now. Some were measured all years, some were measured only a few years (destroyed/newly added plots). The original datasets is split in 2 (deciduous vs coniferous), both containing ~55.000 entries. Per plot about 100 saplings were measured every time, so the two separate datasets probably contain about 50 trees per plot (though this will not always be the case, since the decid:conif ratio is not always equal). Each plot consists of 4 subplots.
I am aware that there might be spatial autocorrelation due to plot placement, but we will not correct for this, yet.
Every year the vegetation is surveyed in the same period. Vegetation cover is estimated at plot-level, individual trees (binary) are measured at a subplot-level.
All trees are measured, so the amount of responses per subplot will differ between subplots and years, as the forest naturally regenerates.
Unfortunately, I cannot share my original data, but I tried to create an example that captures the essentials:
#set seed for whole procedure
addTaskCallback(function(...) {set.seed(453);TRUE})
# Generate vector containing individual vegetation covers (in %)
cover1vec <- c(sample(0:100,10, replace = TRUE)) #the ',number' is amount of covers generated
# Create dataset
DT <- data.frame(
eaten = sample(c(0,1), 80, replace = TRUE),
plot = as.factor(rep(c(1:5), each = 16)),
subplot = as.factor(rep(c(1:4), each = 2)),
year = as.factor(rep(c(2012,2013), each = 8)),
cover1 = rep(cover1vec, each = 8)
)
Which will generate this dataset:
>DT
eaten plot subplot year cover1
1 0 1 1 2012 4
2 0 1 1 2012 4
3 1 1 2 2012 4
4 1 1 2 2012 4
5 0 1 3 2012 4
6 1 1 3 2012 4
7 0 1 4 2012 4
8 1 1 4 2012 4
9 1 1 1 2013 77
10 0 1 1 2013 77
11 0 1 2 2013 77
12 1 1 2 2013 77
13 1 1 3 2013 77
14 0 1 3 2013 77
15 1 1 4 2013 77
16 0 1 4 2013 77
17 0 2 1 2012 46
18 0 2 1 2012 46
19 0 2 2 2012 46
20 1 2 2 2012 46
....etc....
80 0 5 4 2013 82
Note1: to clarify again, in this example the number of responses is the same for every subplot:year combination, making the data balanced, which is not the case in the original dataset.
Note2: this example can not be run in a GLMM, as I get a singularity warning and all my random effect measurements are zero. Apparently my example is not appropriate to actually use (because using sample() caused the 0 and 1 to be in too even amounts to have large enough effects?).
As you can see from the example, cover data is the same for every plot:year combination.
Plots are measured multiple years (only 2012 and 2013 in the example), so there are repeated measures.
Additionally, a year effect is likely, given the fact that we have e.g. drier/wetter years.
First I thought about the following model structure:
library(lme4)
mod1 <- glmer(eaten ~ cover1 + (1 | year) + (1 | plot), data = DT, family = binomial)
summary(mod1)
Where (1 | year) should correct for differences between years and (1 | plot) should correct for the repeated measures.
But then I started thinking: all trees measured in plot 1, during year 2012 will be more similar to each other than when they are compared with (partially the same) trees from plot 1, during year 2013.
So, I doubt that this random model structure will correct for this within plot temporal effect.
So my best guess is to add another random variable, where this "interaction" is accounted for.
I know of two ways to possibly achieve this:
Method 1.
Adding the random variable " + (1 | year:plot)"
Method 2.
Adding the random variable " + (1 | year/plot)"
From what other people told me, I still do not know the difference between the two.
I saw that Method 2 added an extra random variable (year.1) compared to Method 1, but I do not know how to interpret that extra random variable.
As an example, I added the Random effects summary using Method 2 (zeros due to singularity issues with my example data):
Random effects:
Groups Name Variance Std.Dev.
plot.year (Intercept) 0 0
plot (Intercept) 0 0
year (Intercept) 0 0
year.1 (Intercept) 0 0
Number of obs: 80, groups: plot:year, 10; plot, 5; year, 2
Can someone explain me the actual difference between Method 1 and Method 2?
I am trying to understand what is happening, but cannot grasp it.
I already tried to get advice from a colleague and he mentioned that it is likely more appropriate to use cbind(success, failure) per plot:year combination.
Via this site I found that cbind is used in binomial models when Ntrails > 1, which I think is indeed the case given our sampling procedure.
I wonder, if cbind is already used on a plot:year combination, whether I need to add a plot:year random variable?
When using cbind, the example data would look like this:
>DT3
plot year cover1 Eaten_suc Eaten_fail
8 1 2012 4 4 4
16 1 2013 77 4 4
24 2 2012 46 2 6
32 2 2013 26 6 2
40 3 2012 91 2 6
48 3 2013 40 3 5
56 4 2012 61 5 3
64 4 2013 19 2 6
72 5 2012 19 5 3
80 5 2013 82 2 6
What would be the correct random model structure and why?
I was thinking about:
Possibility A
mod4 <- glmer(cbind(Eaten_suc, Eaten_fail) ~ cover1 + (1 | year) + (1 | plot),
data = DT3, family = binomial)
Possibility B
mod5 <- glmer(cbind(Eaten_suc, Eaten_fail) ~ cover1 + (1 | year) + (1 | plot) + (1 | year:plot),
data = DT3, family = binomial)
But doesn't cbind(success, failure) already correct for the year:plot dependence?
Possibility C
mod6 <- glmer(cbind(Eaten_suc, Eaten_fail) ~ cover1 + (1 | year) + (1 | plot) + (1 | year/plot),
data = DT3, family = binomial)
As I do not yet understand the difference between year:plot and year/plot
Thus: Is it indeed more appropriate to use the cbind-method than the raw binary data? And what random model structure would be necessary to prevent pseudoreplication and other dependencies?
Thank you in advance for your time and input!
EDIT 7/12/20: I added some extra information about the original data
You are asking quite a few questions in your question. I'll try to cover them all, but I do suggest reading the documentation and vignette from lme4 and the glmmFAQ page for more information. Also I'd highly recommend searching for these topics on google scholar, as they are fairly well covered.
I'll start somewhere simple
Note 2 (why is my model singular?)
Your model is highly singular, because the way you are simulating your data does not indicate any dependency between the data itself. If you wanted to simulate a binomial model you would use g(eta) = X %*% beta to simulate your linear predictor and thus the probability for success. One can then use this probability for simulating the your binary outcome. This would thus be a 2 step process, first using some known X or randomly simulated X given some prior distribution of our choosing. In the second step we would then use rbinom to simulate binary outcome while keeping it dependent on our predictor X.
In your example you are simulating independent X and a y where the probability is independent of X as well. Thus, when we look at the outcome y the probability of success is equal to p=c for all subgroup for some constant c.
Can someone explain me the actual difference between Method 1 and Method 2? ((1| year:plot) vs (1|year/plot))
This is explained in the package vignette fitting linear mixed effects models with lme4 in the table on page 7.
(1|year/plot) indicates that we have 2 mixed intercept effects, year and plot and plot is nested within year.
(1|year:plot) indicates a single mixed intercept effect, plot nested within year. Eg. we do not include the main effect of year. It would be somewhat similar to having a model without intercept (although less drastic, and interpretation is not destroyed).
It is more common to see the first rather than the second, but we could write the first as a function of the second (1|year) + (1|year:plot).
Thus: Is it indeed more appropriate to use the cbind-method than the raw binary data?
cbind in a formula is used for binomial data (or multivariate analysis), while for binary data we use the raw vector or 0/1 indicating success/failure, eg. aggregate binary data (similar to how we'd use glm). If you are uninterested in the random/fixed effect of subplot, you might be able to aggregate your data across plots, and then it would likely make sense. Otherwise stay with you 0/1 outcome vector indicating either success or failures.
What would be the correct random model structure and why?
This is a topic that is extremely hard to give a definitive answer to, and one that is still actively researched. Depending on your statistical paradigm opinions differ greatly.
Method 1: The classic approach
Classic mixed modelling is based upon knowledge of the data you are working with. In general there are several "rules of thumb" for choosing these parameters. I've gone through a few in my answer here. In general if you are "not interested" in the systematic effect and it can be thought of as a random sample of some population, then it could be a random effect. If it is the population, eg. samples do not change if the process is repeated, then it likely shouldn't.
This approach often yields "decent" choices for those who are new to mixed effect models, but is highly criticized by authors who tend towards methods similar to those we'd use in non-mixed models (eg. visualizing to base our choice and testing for significance).
Method 2: Using visualization
If you are able to split your data into independent subgroups and keeping the fixed effect structure a reasonable approach for checking potential random effects is the estimate marginal models (eg. using glm) across these subgroups and seeing if the fixed effects are "normally distributed" between these observations. The function lmList (in lme4) is designed for this specific approach. In linear models we would indeed expect these to be normally distributed, and thus we can get an indication whether a specific grouping "might" be a valid random effect structure. I believe the same is approximately true in the case of generalized linear models, but I lack references. I know that Ben Bolker have advocated for this approach in a prior article of his (the first reference below) that I used during my thesis. However this is only a valid approach for strictly separable data, and the implementation is not robust in the case where factor levels are not shared across all groups.
So in short: If you have the right data, this approach is simple, fast and seemingly highly reliable.
Method 3: Fitting maximal/minimal models and decreasing/expanding model based on AIC or AICc (or p-value tests or alternative metrics)
Finally an alternative to use a "step-wise"-like procedure. There are advocates of both starting with maximal and minimal models (I'm certain at least one of my references below talk about problems with both, otherwise check glmmFAQ) and then testing your random effects for their validity. Just like classic regression this is somewhat of a double-edged sword. The reason is both extremely simple to understand and amazingly complex to comprehend.
For this method to be successful you'd have to perform cross-validation or out-of-sample validation to avoid selection bias just like standard models, but unlike standard models sampling becomes complicated because:
The fixed effects are conditional on the random structure.
You will need your training and testing samples to be independent
As this is dependent on your random structure, and this is chosen in a step-wise approach it is hard to avoid information leakage in some of your models.
The only certain way to avoid problems here is to define the space
that you will be testing and selecting samples based on the most
restrictive model definition.
Next we also have problems with choice of metrics for evaluation. If one is interested in the random effects it makes sense to use AICc (AIC estimate of the conditional model) while for fixed effects it might make more sense to optimize AIC (AIC estimate of the marginal model). I'd suggest checking references to AIC and AICc on glmmFAQ, and be wary since the large-sample results for these may be uncertain outside a very reestrictive set of mixed models (namely "enough independent samples over random effects").
Another approach here is to use p-values instead of some metric for the procedure. But one should likely be even more wary of test on random effects. Even using a Bayesian approach or bootstrapping with incredibly high number of resamples sometimes these are just not very good. Again we need "enough independent samples over random effects" to ensure the accuracy.
The DHARMA provides some very interesting testing methods for mixed effects that might be better suited. While I was working in the area the author was still (seemingly) developing an article documenting the validity of their chosen method. Even if one does not use it for initial selection I can only recommend checking it out and deciding upon whether one believes in their methods. It is by far the most simple approach for a visual test with simple interpretation (eg. almost no prior knowledge is needed to interpret the plots).
A final note on this method would thus be: It is indeed an approach, but one I would personally not recommend. It requires either extreme care or the author accepting ignorance of model assumptions.
Conclusion
Mixed effect parameter selection is something that is difficult. My experience tells me that mostly a combination of method 1 and 2 are used, while method 3 seems to be used mostly by newer authors and these tend to ignore either out-of-sample error (measure model metrics based on the data used for training), ignore independence of samples problems when fitting random effects or restrict themselves to only using this method for testing fixed effect parameters. All 3 do however have some validity. I myself tend to be in the first group, and base my decision upon my "experience" within the field, rule-of-thumbs and the restrictions of my data.
Your specific problem.
Given your specific problem I would assume a mixed effect structure of (1|year/plot/subplot) would be the correct structure. If you add autoregressive (time-spatial) effects likely year disappears. The reason for this structure is that in geo-analysis and analysis of land plots the classic approach is to include an effect for each plot. If each plot can then further be indexed into subplot it is natural to think of "subplot" to be nested in "plot". Assuming you do not model autoregressive effects I would think of time as random for reasons that you already stated. Some years we'll have more dry and hotter weather than others. As the plots measured will have to be present in a given year, these would be nested in year.
This is what I'd call the maximal model and it might not be feasible depending on your amount of data. In this case I would try using (1|time) + (1|plot/subplot). If both are feasible I would compare these models, either using bootstrapping methods or approximate LRT tests.
Note: It seems not unlikely that (1|time/plot/subplot) would result in "individual level effects". Eg 1 random effect per row in your data. For reasons that I have long since forgotten (but once read) it is not plausible to have individual (also called subject-level) effects in binary mixed models. In this case It might also make sense to use the alternative approach or test whether your model assumptions are kept when withholding subplot from your random effects.
Below I've added some useful references, some of which are directly relevant to the question. In addition check out the glmmFAQ site by Ben Bolker and more.
References
Bolker, B. et al. (2009). „Generalized linear mixed models: a practical guide for ecology and evolution“. In: Trends in ecology & evolution 24.3, p. 127–135.
Bolker, B. et al. (2011). „GLMMs in action: gene-by-environment interaction in total fruit production of wild populations of Arabidopsis thaliana“. In: Revised version, part 1 1, p. 127–135.
Eager, C. og J. Roy (2017). „Mixed effects models are sometimes terrible“. In: arXiv preprint arXiv:1701.04858. url: https://arxiv.org/abs/1701.04858 (last seen 19.09.2019).
Feng, Cindy et al. (2017). „Randomized quantile residuals: an omnibus model diagnostic tool with unified reference distribution“. In: arXiv preprint arXiv:1708.08527. (last seen 19.09.2019).
Gelman, A. og Jennifer Hill (2007). Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press.
Hartig, F. (2019). DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. R package version 0.2.4. url: http://florianhartig.github.io/DHARMa/ (last seen 19.09.2019).
Lee, Y. og J. A. Nelder (2004). „Conditional and Marginal Models: Another View“. In: Statistical Science 19.2, p. 219–238.
doi: 10.1214/088342304000000305. url: https://doi.org/10.1214/088342304000000305
Lin, D. Y. et al. (2002). „Model-checking techniques based on cumulative residuals“. In: Biometrics 58.1, p. 1–12. (last seen 19.09.2019).
Lin, X. (1997). „Variance Component Testing in Generalised Linear Models with Random Effects“. In: Biometrika 84.2, p. 309–326. issn: 00063444. url: http://www.jstor.org/stable/2337459
(last seen 19.09.2019).
Stiratelli, R. et al. (1984). „Random-effects models for serial observations with binary response“. In:
Biometrics, p. 961–971.
I am creating a decision tree with n=3410. The target value contains 6 unique values. Each of these n=3410 have one of these 6 values. The distribution of the values in the data set used to create the model are:
1 - 242
2 - 917
3 - 645
4 - 488
5 - 261
6 - 841
However, when creating the model from this data, values 1 and 5 have a 100% error rate. The root node error rate is also super high - 73%
I'm trying to understand what can cause this problem. I see the relative occurrence of these 2 values in the set is lower, but not statistically insignificant. I can't explain the root node error at all though.
I've tried tuning the tree and manipulating the data set itself, but I am still consistently getting an overall error in the matrix of about 60%. I'm not really understanding what this means or how I can improve it - or if it's just the data I'm using.
I try to figure out how to deal with my forecasting problem and I am not sure if my understanding is right in this field, so it would be really nice if someone can help me. First of all, my goal is to forecast a time series with regression. Instead of using ARIMA model or other heuristic models I want to focus on machine learning techniques like regressions such as random forest regression, k-nearest-neighbour regression etc.. Here is an overview of the dataset:
Timestamp UsageCPU UsageMemory Indicator Delay
2014-01-03 21:50:00 3123 1231 1 123
2014-01-03 22:00:00 5123 2355 1 322
2014-01-03 22:10:00 3121 1233 2 321
2014-01-03 22:20:00 2111 1234 2 211
2014-01-03 22:30:00 1000 2222 2 0
2014-01-03 22:40:00 4754 1599 1 0
The timestamp is increased in steps of 10 minutes and I want to predict the independent variable UsageCPU with the dependent variables UsageMemory, Indicator etc.. At this point i will explain my general knowledge of the prediction part. So for the prediction it is necessary to separate the dataset into training, validation and test sets. For this my dataset that contains 2 whole weeks is separated in 60% training, 20% validation and 20% test. This means for training set I have the first 8 days included and for the validation and the test set I have each 3 days. After that I can train a model in SparkR (the settings are not important).
model <- spark.randomForest(train, UsageMemory ~ UsageMemory, Indicator, Delay,
type = "regression", maxDepth = 30, maxBins = 50, numTrees=50,
impurity="variance", featureSubsetStrategy="all")
So after this I can validate the results with the validation set and compute the RMSE to see the accuracy of the model and which point have to tuned in my model building part. If that is finished I can predict on the test dataset:
predictions <- predict(model, test)
So the prediction works fine, but this is only an in-sample forecast and can not be used to predict for example the next day. In my understanding the in-sample can only used to predict the data in the data set and not to predict future values that can happen tomorrow. So really want to predict for example the next day or only the next 10 minutes / 1 hour, which is only possible to success with the out-of-sample forecasting. I also tried something like this (rolling regression) on the predicted values from random forest, but in my case the rolling regression is only used for evaluating the performance of different regressors with respect to different parameters combinations. So this is in my understanding no out-sample forecasting.
t <- bind(prediction, RollingRegression3 = rollApply(prediction, fun=function(x) mean(UsageCPU), window=6, align='right'))
So in my understanding I need something (maybe lag values?), before the model building process starts. I also read a lot of different papers and books, but there is no clear way how to do it and what are the key points. There is only standing something like t+1, t+n, but right now I do not even know how to do it. Would be really nice if someone can help me, because I tried to figure this out since three month now, thank you.
Let’s see if I get your problem right. I suppose that, given a time window, e.g. 144 last observations (one day) of UsageCPU, UsageMemory, Indicator and Delay, you want to forecast the ‘n’ next observations of UsageCPU. One way you could do such a thing, using random forests, is assigning one model for each next observation you want to forecast. So, if you want to forecast the 10 next UsageCPU observations, you should train 10 random forest models.
Using the example I began with, you could split the data you have in chunks of 154 observations. In each, you will use the first 144 observations to forecast the last 10 values of UsageCPU. There are lots of ways in which you could use feature engineering to extract information from these first 144 observations to train your model with, e.g. mean for each variable, last observation of each variable, global mean for each variable. So, for each chunk you will get a vector containing a bunch of predictors and 10 target values.
Bind the vectors you got for each chunk and you’ll have a matrix where the first columns are the predictors and the last 10 columns are the targets. Train each random forest with the n predictors columns and 1 of the targets column. Now you can apply the models on the features you extract from any data chunk containing the 144 observations. The model trained for target column 1 will ‘forecast’ one observation ahead, the model trained for target column 2 will ‘forecast’ two observations ahead, the model trained for target column 3 will ‘forecast’ three observations ahead...