mlr3 | mlr3 : Machine Learning in R - next generation | Machine Learning library

I have the following problem. I want to build a model for landcover classification. My data are multitemporal Remote Sensing data with several bands. For training I created stratified randomly distributed points to extract spectral data at their positions. With these data a Random Forrest (Rpart) was trained using mlr3 package. For accuracy measurement a repeated spatial cross validation using mlr3spatiotempcv was performed. The resulting model of the training step is, after extraction, stored in an R Object of type rpart. In the terms field of this object are the variable names stored. These are all my used bands but also the spatial x and y coordinates. This brings problems when predicting new data. I used terra package and got an error the x and y layer are missing in my input data. Which kind of makes sense because they are stored in the terms field of the model. But from my understanding, the coordinates should not be a variable of the model. The coordinates are just used for spatial resampling and not for predicting. I "solved" this problem by removing x and y coordinates during the training process and perform just an ordinary non-spatial cross validation. After that I performed the prediction and it works perfectly.

So, my Question is, how can I train a model, using mlr3 package, with data containing coordinates, to perform spatial cross validation?, and then use this model to predict a new Raster.

How can I control the number of configurations being evaluated during hyperband tuning in mlr3? I noticed that when I tune 6 parameters in xgboost(), the code evaluates about 9 configurations. When I tune the same number of parameters in catboost(), the code starts with evaluating 729 configurations. I am using eta = 3 in both cases.

I want to code some hyperband tuning in mlr3. I started with running the subsample-rpart hyperband example from chapter 4.4 in mlr3 book - directly copied from there. I am getting an error: Error in benchmark(design, store_models = self$store_models, allow_hotstart = self$allow_hotstart, : unused argument (clone = character())

How do I fix it?

I am a beginner on mlr3 and am facing problems while running AutoFSelector learner associated to glmnet on a classification task containing >2000 numeric variables. I reproduce this error while using the simpler mlr3 predefined task Sonar. For note, I am using R version 4.1.2 (2021-11-01)on macOS Monterey 12.1. All required packages have been loaded on CRAN.

I have a task where the observation in rows have a date order. I generate a custom resampling scheme that respects this order in all train/test splits.

I also want to adress the unbalanced classes problem by upsampling the minority class. Within the training sets, the time order is not important (and the learner would not use this anyway).

Now, I want to resample this combination of ordered task, graph learner (including upsampling) and time-sensitive custom resampling scheme. But this is problematic.

To show this, I generated the following code. I use a sample task to make this reproducible and I augment this task with a date column to generate an ordered task that is similar to my problem. This code runs only if I omit the problematic lines indicated in the code. But they generate exactly what I have in my real-world problem: An order. So how can I solve this?

(I omit some of the output in the following reprex for readability.)

The code below is the sample from the mlr3cluster Github repo. My question is whether the learner is overtrained, due to the task not being split into a train and test set, or does it take care of that on its own internally?

My guess is that it does but I'm just not sure and I'm new to R and mlr3 and can't seem to find documentation regarding this topic.

I am using the benchmark() function in mlr3 to compare several ML algorithms. One of them is XGB with hyperparameter tuning. Thus, I have an outer resampling to evaluate the overall performance (hold-out sample) and an inner resampling for the hyper parameter tuning (5-fold Cross-validation). Besides having an estimate of the accuracy for all ML algorithms, I would like to see the feature importance of the tuned XGB. For that, I would have to access the tuned model (within the benchmark object). I do not know how to do that. The object returned by benchmark() is a deeply nested list and I do not understand its structure.

This answer on stackoverflow did not help me, because it uses a different setup (a learner in a pipeline rather than a benchmark object).

This answer on github did not help me, because it shows how to extract all the information about the benchmarking at once but not how to extract one (tuned) model of one of the learners in the benchmark.

Below is the code I am using to carry out the nested resampling. Following the benchmarking, I would like to estimate the feature importance as described here, which requires accessing the tuned XGB model.

I have some following codes. I met error when save trained model. It's only error when i using lightgbm.

I want to run a number of machine learning algorithms with different feature selection methods on survival data using the MLR3 package. For that, I am using the Benchmark() function of MLR3.

Unfortunately, filter feature selection methods of MLR3 do not support survival, yet. However, MLR package supports survival filters.

I can fuse MLR learners with an MLR filter method. After that, I need to convert them to a learner in MLR3 in order to be able to use banchmark_grid() function of MLR3.

Is there any way to use MLR survival filters in MLR3? Or is there any way that I can convert MLR filters to MLR3 filters?