Anthony Blaom, PhD
Anthony Blaom, PhD
[mlflow](https://www.mlflow.org) is an API and web-based UI for logging parameters, code versions, metrics, and output files when running machine learning experiments, for later visualizing the results. Integration of mlflow already...
Currently, `evaluate!(mach, ...)` will mutate `mach` because `mach` is repeatedly retrained on different views of the data; `mach` is never trained on *all* supplied data (or all specified `rows=...`). But...
https://discourse.julialang.org/t/time-normalisation-results-in-nothing-gradients-of-ode-solutions/109353/3
As we can see from the errors below, this issue is about having a general fallback to get the number of rows of a table (where known) which has come...
These include `MatrixTable`, `DictRowTable`, `DictColumnTable`, but perhaps there are others I don't know about. In support of: #330
The following works fine if `outdim=0` is replaced with `outdim=3`. A more informative error message would be helpful, unless a better inference of `outdim` is more appropriate. Sorry this is...
As far as I can tell, the `loss` parameter is only exposed for single trees. I think this would be pretty easy to add to the ensemble models. Issue raised...
Here's what I think are intended to be public, ranked in rough order of priority: - [x] `build_forest` #214 @ablaom - [x] `apply_forest` - [x] `apply_forest_proba` - [ ] `build_tree`...
When multithreading prediction support was added in https://github.com/JuliaAI/DecisionTree.jl/pull/188/files it appears that it was only added for `apply_forest` and not `apply_forest_proba`. And not added at all for regression, only classification.
When you want to train on all specified data and test on the same data. So, ```julia e = evaluate(model, X, y, resampling=InSample(), measure=l1) e.measurement[1] ``` is equivalent to: ```julia...