Xgboost implement Aft and Cox with weights, untested need support for testing #163
Conversation
|
#161 I will try to implement on this branch |
|
We also may need to create a dummy metric for censored regression models in order to leverage the stacks framework, or just reimplement it in censored knowing that the components won't be in individually tuned but the stacked predictor can be tuned because it can calculate some metrics like c-index and brier score, something the base models can't do |
|
I think I got close to what I wanted to achieve, maybe we can create a simpler blend_predictions for survival models because as the documentation clearly show not all models will be able to produce all kinds of prediction types, but we can stack them as long as there are non time dependent predictions. library(tidymodels)
#> Registered S3 method overwritten by 'tune':
#> method from
#> required_pkgs.model_spec parsnip
library(censored)
#> Loading required package: survival
library(survival)
tidymodels_prefer()
data(cancer)
lung <- lung %>% drop_na()
lung_train <- lung[-c(1:5), ]
lung_test <- lung[1:5, ]
mod_spec1 <-
decision_tree() %>%
set_mode("censored regression")
mod_spec2 <-
boost_tree(trees = 100) %>%
set_mode("censored regression") |>
set_engine('xgboost',aft_loss_distribution_scale = 1)
models <- list(mod_spec1,mod_spec2)
fits <- models |> map(fit,data = lung_train,formula = Surv(time, status) ~ .)
table_specs <- tibble(fits)
table_predicts <- table_specs |>
rowwise() |>
mutate(predicts = list(predict(fits, lung_train)))
train_predictions_models <- table_predicts |>
ungroup() |>
select(predicts) |>
mutate(model_number = stringr::str_c('model_', row_number())) |>
unnest(cols = predicts) |>
pivot_wider(
names_from = model_number,
values_from = -model_number,
values_fn = list
) |> unnest(everything())
# create the train stacked model train
stacked_data <- lung_train |>
select(time,status) |>
bind_cols(train_predictions_models)
spec_stack <-
proportional_hazards(penalty = .1) |>
set_engine("glmnet",
lower.limits = 0, lambda.min.ratio = 0)
fit_stacked <- spec_stack |> fit(Surv(time, status) ~ .,data = stacked_data)
fit_stacked |> predict(stacked_data,type = 'survival',time = c(1,10,100,200)) |>
slice(119) |>
tidyr::unnest(col = .pred)
#> # A tibble: 4 x 2
#> .time .pred_survival
#> <dbl> <dbl>
#> 1 1 1
#> 2 10 0.995
#> 3 100 0.858
#> 4 200 0.689
table_predicts_testing <- table_predicts |>
mutate(predict_testing = list(predict(fits, lung_test)))
test_predictions_models <- table_predicts_testing |>
ungroup() |>
select(predict_testing) |>
mutate(model_number = stringr::str_c('model_', row_number())) |>
unnest(cols = predict_testing ) |>
pivot_wider(
names_from = model_number,
values_from = -model_number,
values_fn = list
) |> unnest(everything())
stacked_data_testing <- lung_test |>
select(time,status) |>
bind_cols(test_predictions_models)
fit_stacked |> predict(stacked_data_testing,type = 'survival',time = c(1,10,100,200)) |>
slice(1:5) |>
tidyr::unnest(col = .pred)
#> # A tibble: 20 x 2
#> .time .pred_survival
#> <dbl> <dbl>
#> 1 1 1
#> 2 10 0.995
#> 3 100 0.858
#> 4 200 0.689
#> 5 1 1
#> 6 10 0.993
#> 7 100 0.803
#> 8 200 0.587
#> 9 1 1
#> 10 10 0.997
#> 11 100 0.910
#> 12 200 0.795
#> 13 1 1
#> 14 10 0.994
#> 15 100 0.835
#> 16 200 0.646
#> 17 1 1
#> 18 10 0.994
#> 19 100 0.835
#> 20 200 0.646Created on 2022-03-17 by the reprex package (v2.0.1) |
There was a problem hiding this comment.
Thanks for this! I left some comments but generally I think we should export/use some stuff from parsnip so that we don't duplicate efforts.
Can you elaborate a bit what you mean with your comment about the hazard? And can you add tests for the new functionality as well, please?
R/boost_tree_data_xgboost.R
Outdated
| eng = "xgboost", | ||
| mode = "censored regression", | ||
| options = list( | ||
| predictor_indicators = "none", |
There was a problem hiding this comment.
I think we should leave this as "one_hot" like in parsnip
There was a problem hiding this comment.
The cox objective from xgboost says:
survival:cox: Cox regression for right censored survival time data (negative values are considered right censored). Note that predictions are returned on the hazard ratio scale (i.e., as HR = exp(marginal_prediction) in the proportional hazard function h(t) = h0(t) * HR).
But the aft objective says
survival:aft: Accelerated failure time model for censored survival time data. See Survival Analysis with Accelerated Failure Time for details. and returns time
R/boost_tree_data_xgboost.R
Outdated
|
|
||
|
|
||
|
|
||
| xgb_pred <- function(object, newdata, ...) { |
There was a problem hiding this comment.
I'll make a PR to parsnip to export xgb_pred() from there so we can use it here and don't need a copy
There was a problem hiding this comment.
this is now exported in the dev version of parsnip as xgb_predict()
R/xgb_train_censored.R
Outdated
|
|
||
|
|
||
|
|
||
| as_xgb_data <- function(x, y, validation = 0, ...) { |
There was a problem hiding this comment.
It would be helpful to develop a version that works for all modes here, and when it's ready, open a PR on parsnip. Could you extend the parsnip version?
There was a problem hiding this comment.
maybe... we would need to check the current objective, that was my initial idea to implement aft on parsnip directly, but if we adapt just as_xgb_data the rest should work rather well
There was a problem hiding this comment.
ok, great! let's keep the development of changes to as_xgb_data() here for now though so we don't have to coordinate across multiple PRs.
R/boost_tree_data_xgboost.R
Outdated
| value = list( | ||
| interface = "matrix", | ||
| protect = c("x", "y"), | ||
| func = c(pkg = "censored", fun = "xgb_train_censored"), |
There was a problem hiding this comment.
xgb_train() is exported from parsnip, can you use that with objective = "survival:aft"? you should be able to set that in the defaults in the line below and also protect it in the line above.
There was a problem hiding this comment.
Oh that is genius, know I understand what protect does!, ok makes a lot o sense, just need to get as_xgb_data to work with cox and aft, I can pr this on this week no problem, it shouldn't mess up parsnip since I will do an if statement depending on the objective, and parsnip already has defaults for objectives
There was a problem hiding this comment.
What kind of labels do you need to set for survival:cox? Different than for survival:aft?
There was a problem hiding this comment.
Yes, xgboost decided that for cox you only need 'label' and you set uncensored to - and censored to +, so 50+ becomes 50 and 100 becomes -100
|
Ah, I think I understand better now. The resulting hazards predictions for the Cox model would be a different prediction type in censored ( Using the Also just for clarification, censored only contains code for model engines (for fitting and predicting). Metrics and stacking and so on go into different places. We have plans to extend support for survival models across all necesasry parts of tidymodels but are currently focussed on the model definition/fitting/predicting part. |
|
I like the idea of ensembling terrible base models into half-decent ones with a "stable" model at the end, but I can see that would be harder to implement, I agree with you, we can focus on implementing aft, I just don't want to restrict it too much and we should think about cox a little bit so we don't end up creating two functions, I also think we have the opportunity to set up |
I couldn't get protect to work on fit time because I won't know which prediction someone wants to do until they predict it, also I called the cox predictions as linear_preds's but I am pretty sure that is not right
|
I have updated the function to do aft and cox, I also check to see if it is the right objective library(tidymodels)
#> Registered S3 method overwritten by 'tune':
#> method from
#> required_pkgs.model_spec parsnip
library(censored)
#> Carregando pacotes exigidos: survival
library(survival)
tidymodels_prefer()
data(cancer)
lung <- lung %>% drop_na()
lung_train <- lung[-c(1:5), ]
lung_test <- lung[1:5, ]
test <-
boost_tree()|> set_engine('xgboost',objective = 'survival:cox') |> set_mode('censored regression')
test |>
translate()
#> Boosted Tree Model Specification (censored regression)
#>
#> Engine-Specific Arguments:
#> objective = survival:cox
#>
#> Computational engine: xgboost
#>
#> Model fit template:
#> censored::xgb_train_censored(x = missing_arg(), y = missing_arg(),
#> objective = "survival:cox", nthread = 1, verbose = 0)
set.seed(1)
bt_fit <- test %>% fit(Surv(time, status) ~ ., data = lung_train)
bt_fit
#> parsnip model object
#>
#> ##### xgb.Booster
#> raw: 55.4 Kb
#> call:
#> xgboost::xgb.train(params = list(eta = 0.3, max_depth = 6, gamma = 0,
#> colsample_bytree = 1, colsample_bynode = 1, min_child_weight = 1,
#> subsample = 1, objective = "survival:cox"), data = x$data,
#> nrounds = 15, watchlist = x$watchlist, verbose = 0, nthread = 1)
#> params (as set within xgb.train):
#> eta = "0.3", max_depth = "6", gamma = "0", colsample_bytree = "1", colsample_bynode = "1", min_child_weight = "1", subsample = "1", objective = "survival:cox", nthread = "1", validate_parameters = "TRUE"
#> xgb.attributes:
#> niter
#> callbacks:
#> cb.evaluation.log()
#> # of features: 8
#> niter: 15
#> nfeatures : 8
#> evaluation_log:
#> iter training_cox_nloglik
#> 1 3.967138
#> 2 3.835229
#> ---
#> 14 3.163877
#> 15 3.109603
predict(
bt_fit,
lung_train,
type = "linear_pred",
)
#> # A tibble: 162 x 1
#> .pred_linear_pred
#> <dbl>
#> 1 1.57
#> 2 3.68
#> 3 0.552
#> 4 0.304
#> 5 0.243
#> 6 0.159
#> 7 15.7
#> 8 0.631
#> 9 2.89
#> 10 0.517
#> # ... with 152 more rows
predict(
bt_fit,
lung_train,
type = "time",
)
#> Error:
#> ! The objective should be survival:aft not survival:coxCreated on 2022-03-24 by the reprex package (v2.0.1) |
|
@brunocarlin You might want to look at the changes in the xib functions in parsnip that are being updated for case weights: tidymodels/parsnip@2100e8f It will be helpful when we merge this if you use some of that restructuring. |
|
@topepo got it, should I implement case weights for survival cases as well? Or just use the new naming and put an error if the user passes weights? |
|
@brunocarlin I added a PR with the case weight changes to Let me know if you want a hand with this part; unfortunately library(survival)
cens_data <- with(aml[1:5,], Surv(time, status))
# Looks like a vector:
cens_data
#> [1] 9 13 13+ 18 23
# but is a hidden matrix:
length(cens_data) # <- as if it is a vector
#> [1] 5
dim(cens_data) # <- this object is sus
#> [1] 5 2
# Be careful about subsetting
cens_data[1:3]
#> [1] 9 13 13+
cens_data[1:3, ]
#> [1] 9 13 13+
cens_data[1:3, 1]
#> [1] 9 13 13
cens_data[1:3, 2] # <- these are different when you have interval censoring
#> [1] 1 1 0
# is it a surv object?
inherits(cens_data, "Surv")
#> [1] TRUE
# is it numeric?
inherits(cens_data, "numeric")
#> [1] FALSE
# but...
is.numeric(cens_data)
#> [1] TRUE
# ¯\_(ツ)_/¯Created on 2022-04-04 by the reprex package (v2.0.1) |
|
@brunocarlin In regard to case weights, the main changes are in tidymodels/parsnip#696 so no need to do anything else for that. The won't work until tidymodels/parsnip#692 is merged (which will be a while) |
made some changes in order to look more like the proposed new naming scheme
fixed some bugs on the first update
|
@topepo I think I got the gist of the naming scheme, I have opted into using train instead of training for some variables is that ok? |
There was a problem hiding this comment.
@brunocarlin If you want to do both model types then let's make them two separate engines: xgboost_aft and xgboost_cox. That fits much better with the design of parsnip where a combination of model spec (here: boost_tree()) and engine should only fit one model type. So let's protect the objective arg and remove the pre hooks in the prediction modules. Since each model (AFT and Cox) can only provide one type of prediction, they should both only have one prediction module each (the thing you set with set_pred()). Could you also formalize your testing in unit tests, please?
|
@hfrick Ok, I will try to work on it this week |
I keep getting Error in inherits(new_data, "xgb.DMatrix") : argument "new_data" is missing, with no default
|
@topepo and @hfrick Hey guys, I am back, sorry for the long delay I have tried splitting the engines but when importing the new xgb_predict I get the error library(tidymodels)
library(censored)
#> Loading required package: survival
library(survival)
tidymodels_prefer()
data(cancer)
lung <- lung %>% drop_na()
lung_train <- lung[-c(1:5), ]
lung_test <- lung[1:5, ]
test <-
boost_tree()|> set_engine('xgboost_aft') |> set_mode('censored regression')
test |>
translate()
#> Boosted Tree Model Specification (censored regression)
#>
#> Computational engine: xgboost_aft
#>
#> Model fit template:
#> censored::xgb_train_censored(x = missing_arg(), y = missing_arg(),
#> objective = missing_arg(), nthread = 1, verbose = 0, objective = "survival:aft")
set.seed(1)
bt_fit <- test %>% fit(Surv(time, status) ~ ., data = lung_train)
bt_fit
#> parsnip model object
#>
#> ##### xgb.Booster
#> raw: 36.3 Kb
#> call:
#> xgboost::xgb.train(params = list(eta = 0.3, max_depth = 6, gamma = 0,
#> colsample_bytree = 1, colsample_bynode = 1, min_child_weight = 1,
#> subsample = 1, objective = "survival:aft"), data = x$data,
#> nrounds = 15, watchlist = x$watchlist, verbose = 0, nthread = 1)
#> params (as set within xgb.train):
#> eta = "0.3", max_depth = "6", gamma = "0", colsample_bytree = "1", colsample_bynode = "1", min_child_weight = "1", subsample = "1", objective = "survival:aft", nthread = "1", validate_parameters = "TRUE"
#> xgb.attributes:
#> niter
#> callbacks:
#> cb.evaluation.log()
#> # of features: 8
#> niter: 15
#> nfeatures : 8
#> evaluation_log:
#> iter training_aft_nloglik
#> 1 14.698676
#> 2 9.883017
#> ---
#> 14 4.564666
#> 15 4.553461
predict(
bt_fit,
lung_test,
type = "linear_pred",
)
#> Error in `check_spec_pred_type()`:
#> ! No linear_pred prediction method available for this model.
#> • Value for `type` should be one of: 'time'
#> Backtrace:
#> ▆
#> 1. ├─stats::predict(bt_fit, lung_test, type = "linear_pred", )
#> 2. └─parsnip::predict.model_fit(...)
#> 3. ├─parsnip::predict_linear_pred(...)
#> 4. └─parsnip::predict_linear_pred.model_fit(...)
#> 5. └─parsnip:::check_spec_pred_type(object, "linear_pred")
#> 6. └─rlang::abort(...)
predict(
bt_fit,
lung_test,
type = 'time'
)
#> Error in inherits(new_data, "xgb.DMatrix"): argument "new_data" is missing, with no defaultCreated on 2022-08-27 with reprex v2.0.2 |
|
Hey @brunocarlin! Apologies that it has taken a while to get back to you! I've been meaning to return to this PR but haven't found the bandwidth for it. We have now decided that we'll focus on advancing support for censored regression in other places of tidymodels. This means that I'll take a deeper look into debugging the error in this branch later, most likely when I return to adding additional engines to censored. I appreciate you getting this started and the various pointers to the relevant docs of xgboost! |
|
note to self: add any additional learnings from this to the checklist for adding an engine to what is currently here: tidymodels/tidymodels#97 |
|
@hfrick No problems, happy to help on this as I also learned a lot |
brunocarlin
changed the title
|
After all these years lol, but anyways @hfrick this should cover the basic functionality for aft and cox models that the xgboost package provides out of the box, I have just basically wrapped the data back and forth with the right format on the dread Xgboost Matrix, it should work out of the box according to my own testing, but I haven't stress tested arguments, although I don't see why it shouldn't work with weights and arguments. Also I used the version I had on hand for censored, maybe something else changed for .3 library(tidymodels)
library(censored)
#> Carregando pacotes exigidos: survivallibrary(survival)
tidymodels_prefer()
data(cancer)
lung <- lung %>% drop_na()
lung_train <- lung[-c(1:5), ]
lung_test <- lung[1:5, ]
# Cox
test <-
boost_tree()|> set_engine('xgboost_cox') |> set_mode('censored regression')
test |>
translate()
#> Boosted Tree Model Specification (censored regression)
#>
#> Computational engine: xgboost_cox
#>
#> Model fit template:
#> censored::xgb_train_censored(x = missing_arg(), y = missing_arg(),
#> objective = missing_arg(), nthread = 1, verbose = 0, objective = "survival:cox")set.seed(1)
bt_fit <- test %>% fit(Surv(time, status) ~ ., data = lung_train)
bt_fit
#> parsnip model object
#>
#> ##### xgb.Booster
#> raw: 40.5 Kb
#> call:
#> xgboost::xgb.train(params = list(eta = 0.3, max_depth = 6, gamma = 0,
#> colsample_bytree = 1, colsample_bynode = 1, min_child_weight = 1,
#> subsample = 1, objective = "survival:cox"), data = x$data,
#> nrounds = 15, watchlist = x$watchlist, verbose = 0, nthread = 1)
#> params (as set within xgb.train):
#> eta = "0.3", max_depth = "6", gamma = "0", colsample_bytree = "1", colsample_bynode = "1", min_child_weight = "1", subsample = "1", objective = "survival:cox", nthread = "1", validate_parameters = "TRUE"
#> xgb.attributes:
#> niter
#> callbacks:
#> cb.evaluation.log()
#> # of features: 8
#> niter: 15
#> nfeatures : 8
#> evaluation_log:
#> iter training_cox_nloglik
#> <num> <num>
#> 1 3.967019
#> 2 3.840237
#> ---
#> 14 3.095573
#> 15 3.054746predict(
bt_fit,
lung_test,
type = "linear_pred",
)
#> # A tibble: 5 × 1
#> .pred_linear_pred
#> <dbl>
#> 1 0.351
#> 2 4.41
#> 3 2.23
#> 4 4.50
#> 5 2.36predict(
bt_fit,
lung_test,
type = 'time'
)
#> Error in `check_spec_pred_type()`:
#> ! No time prediction method available for this model.
#> • Value for `type` should be one of: 'linear_pred'# Aft
test <-
boost_tree()|> set_engine('xgboost_aft') |> set_mode('censored regression')
test |>
translate()
#> Boosted Tree Model Specification (censored regression)
#>
#> Computational engine: xgboost_aft
#>
#> Model fit template:
#> censored::xgb_train_censored(x = missing_arg(), y = missing_arg(),
#> objective = missing_arg(), nthread = 1, verbose = 0, objective = "survival:aft")set.seed(1)
bt_fit <- test %>% fit(Surv(time, status) ~ ., data = lung_train)
bt_fit
#> parsnip model object
#>
#> ##### xgb.Booster
#> raw: 36.3 Kb
#> call:
#> xgboost::xgb.train(params = list(eta = 0.3, max_depth = 6, gamma = 0,
#> colsample_bytree = 1, colsample_bynode = 1, min_child_weight = 1,
#> subsample = 1, objective = "survival:aft"), data = x$data,
#> nrounds = 15, watchlist = x$watchlist, verbose = 0, nthread = 1)
#> params (as set within xgb.train):
#> eta = "0.3", max_depth = "6", gamma = "0", colsample_bytree = "1", colsample_bynode = "1", min_child_weight = "1", subsample = "1", objective = "survival:aft", nthread = "1", validate_parameters = "TRUE"
#> xgb.attributes:
#> niter
#> callbacks:
#> cb.evaluation.log()
#> # of features: 8
#> niter: 15
#> nfeatures : 8
#> evaluation_log:
#> iter training_aft_nloglik
#> <num> <num>
#> 1 14.698676
#> 2 9.883017
#> ---
#> 14 4.564666
#> 15 4.553461predict(
bt_fit,
lung_test,
type = "linear_pred",
)
#> Error in `check_spec_pred_type()`:
#> ! No linear_pred prediction method available for this model.
#> • Value for `type` should be one of: 'time'predict(
bt_fit,
lung_test,
type = 'time'
)
#> # A tibble: 5 × 1
#> .pred_time
#> <dbl>
#> 1 420.
#> 2 239.
#> 3 120.
#> 4 78.7
#> 5 350.Created on 2024-07-04 with reprex v2.1.0 Session infosessioninfo::session_info()
#> ─ Session info ───────────────────────────────────────────────────────────────
#> setting value
#> version R version 4.4.1 (2024-06-14 ucrt)
#> os Windows 11 x64 (build 22631)
#> system x86_64, mingw32
#> ui RTerm
#> language (EN)
#> collate Portuguese_Brazil.utf8
#> ctype Portuguese_Brazil.utf8
#> tz America/Sao_Paulo
#> date 2024-07-04
#> pandoc 3.1.11 @ C:/Program Files/RStudio/resources/app/bin/quarto/bin/tools/ (via rmarkdown)
#>
#> ─ Packages ───────────────────────────────────────────────────────────────────
#> package * version date (UTC) lib source
#> backports 1.5.0 2024-05-23 [1] CRAN (R 4.4.0)
#> broom * 1.0.6 2024-05-17 [1] CRAN (R 4.4.1)
#> cachem 1.1.0 2024-05-16 [1] CRAN (R 4.4.1)
#> censored * 0.0.0.9000 2024-07-04 [1] local
#> class 7.3-22 2023-05-03 [2] CRAN (R 4.4.1)
#> cli 3.6.3 2024-06-21 [1] CRAN (R 4.4.1)
#> codetools 0.2-20 2024-03-31 [2] CRAN (R 4.4.1)
#> colorspace 2.1-0 2023-01-23 [1] CRAN (R 4.4.1)
#> conflicted 1.2.0 2023-02-01 [1] CRAN (R 4.4.1)
#> data.table 1.15.4 2024-03-30 [1] CRAN (R 4.4.1)
#> dials * 1.2.1 2024-02-22 [1] CRAN (R 4.4.1)
#> DiceDesign 1.10 2023-12-07 [1] CRAN (R 4.4.1)
#> digest 0.6.36 2024-06-23 [1] CRAN (R 4.4.1)
#> dplyr * 1.1.4 2023-11-17 [1] CRAN (R 4.4.1)
#> evaluate 0.24.0 2024-06-10 [1] CRAN (R 4.4.1)
#> fansi 1.0.6 2023-12-08 [1] CRAN (R 4.4.1)
#> fastmap 1.2.0 2024-05-15 [1] CRAN (R 4.4.1)
#> foreach 1.5.2 2022-02-02 [1] CRAN (R 4.4.1)
#> fs 1.6.4 2024-04-25 [1] CRAN (R 4.4.1)
#> furrr 0.3.1 2022-08-15 [1] CRAN (R 4.4.1)
#> future 1.33.2 2024-03-26 [1] CRAN (R 4.4.1)
#> future.apply 1.11.2 2024-03-28 [1] CRAN (R 4.4.1)
#> generics 0.1.3 2022-07-05 [1] CRAN (R 4.4.1)
#> ggplot2 * 3.5.1 2024-04-23 [1] CRAN (R 4.4.1)
#> globals 0.16.3 2024-03-08 [1] CRAN (R 4.4.0)
#> glue 1.7.0 2024-01-09 [1] CRAN (R 4.4.1)
#> gower 1.0.1 2022-12-22 [1] CRAN (R 4.4.0)
#> GPfit 1.0-8 2019-02-08 [1] CRAN (R 4.4.1)
#> gtable 0.3.5 2024-04-22 [1] CRAN (R 4.4.1)
#> hardhat 1.4.0 2024-06-02 [1] CRAN (R 4.4.1)
#> htmltools 0.5.8.1 2024-04-04 [1] CRAN (R 4.4.1)
#> infer * 1.0.7 2024-03-25 [1] CRAN (R 4.4.1)
#> ipred 0.9-14 2023-03-09 [1] CRAN (R 4.4.1)
#> iterators 1.0.14 2022-02-05 [1] CRAN (R 4.4.1)
#> jsonlite 1.8.8 2023-12-04 [1] CRAN (R 4.4.1)
#> knitr 1.47 2024-05-29 [1] CRAN (R 4.4.1)
#> lattice 0.22-6 2024-03-20 [2] CRAN (R 4.4.1)
#> lava 1.8.0 2024-03-05 [1] CRAN (R 4.4.1)
#> lhs 1.2.0 2024-06-30 [1] CRAN (R 4.4.1)
#> lifecycle 1.0.4 2023-11-07 [1] CRAN (R 4.4.1)
#> listenv 0.9.1 2024-01-29 [1] CRAN (R 4.4.1)
#> lubridate 1.9.3 2023-09-27 [1] CRAN (R 4.4.1)
#> magrittr 2.0.3 2022-03-30 [1] CRAN (R 4.4.1)
#> MASS 7.3-60.2 2024-04-26 [2] CRAN (R 4.4.1)
#> Matrix 1.7-0 2024-04-26 [2] CRAN (R 4.4.1)
#> memoise 2.0.1 2021-11-26 [1] CRAN (R 4.4.1)
#> modeldata * 1.4.0 2024-06-19 [1] CRAN (R 4.4.1)
#> munsell 0.5.1 2024-04-01 [1] CRAN (R 4.4.1)
#> nnet 7.3-19 2023-05-03 [2] CRAN (R 4.4.1)
#> parallelly 1.37.1 2024-02-29 [1] CRAN (R 4.4.0)
#> parsnip * 1.2.1 2024-03-22 [1] CRAN (R 4.4.1)
#> pillar 1.9.0 2023-03-22 [1] CRAN (R 4.4.1)
#> pkgconfig 2.0.3 2019-09-22 [1] CRAN (R 4.4.1)
#> prodlim 2024.06.25 2024-06-24 [1] CRAN (R 4.4.1)
#> purrr * 1.0.2 2023-08-10 [1] CRAN (R 4.4.1)
#> R.cache 0.16.0 2022-07-21 [1] CRAN (R 4.4.1)
#> R.methodsS3 1.8.2 2022-06-13 [1] CRAN (R 4.4.0)
#> R.oo 1.26.0 2024-01-24 [1] CRAN (R 4.4.0)
#> R.utils 2.12.3 2023-11-18 [1] CRAN (R 4.4.1)
#> R6 2.5.1 2021-08-19 [1] CRAN (R 4.4.1)
#> Rcpp 1.0.12 2024-01-09 [1] CRAN (R 4.4.1)
#> recipes * 1.0.10 2024-02-18 [1] CRAN (R 4.4.1)
#> reprex 2.1.0 2024-01-11 [1] CRAN (R 4.4.1)
#> rlang 1.1.4 2024-06-04 [1] CRAN (R 4.4.1)
#> rmarkdown 2.27 2024-05-17 [1] CRAN (R 4.4.1)
#> rpart 4.1.23 2023-12-05 [2] CRAN (R 4.4.1)
#> rsample * 1.2.1 2024-03-25 [1] CRAN (R 4.4.1)
#> rstudioapi 0.16.0 2024-03-24 [1] CRAN (R 4.4.1)
#> scales * 1.3.0 2023-11-28 [1] CRAN (R 4.4.1)
#> sessioninfo 1.2.2 2021-12-06 [1] CRAN (R 4.4.1)
#> styler 1.10.3 2024-04-07 [1] CRAN (R 4.4.1)
#> survival * 3.7-0 2024-06-05 [1] CRAN (R 4.4.1)
#> tibble * 3.2.1 2023-03-20 [1] CRAN (R 4.4.1)
#> tidymodels * 1.2.0 2024-03-25 [1] CRAN (R 4.4.1)
#> tidyr * 1.3.1 2024-01-24 [1] CRAN (R 4.4.1)
#> tidyselect 1.2.1 2024-03-11 [1] CRAN (R 4.4.1)
#> timechange 0.3.0 2024-01-18 [1] CRAN (R 4.4.1)
#> timeDate 4032.109 2023-12-14 [1] CRAN (R 4.4.1)
#> tune * 1.2.1 2024-04-18 [1] CRAN (R 4.4.1)
#> utf8 1.2.4 2023-10-22 [1] CRAN (R 4.4.1)
#> vctrs 0.6.5 2023-12-01 [1] CRAN (R 4.4.1)
#> withr 3.0.0 2024-01-16 [1] CRAN (R 4.4.1)
#> workflows * 1.1.4 2024-02-19 [1] CRAN (R 4.4.1)
#> workflowsets * 1.1.0 2024-03-21 [1] CRAN (R 4.4.1)
#> xfun 0.45 2024-06-16 [1] CRAN (R 4.4.1)
#> xgboost 1.7.7.1 2024-01-25 [1] CRAN (R 4.4.1)
#> yaml 2.3.8 2023-12-11 [1] CRAN (R 4.4.1)
#> yardstick * 1.3.1 2024-03-21 [1] CRAN (R 4.4.1)
#>
#> [1] C:/Users/bruno/AppData/Local/R/win-library/4.4
#> [2] C:/Program Files/R/R-4.4.1/library
#>
#> ────────────────────────────────────────────────────────────────────────────── |
Very messy start to the functionality, I couldn't remove this gitnore sorry, also we will eventually have to implement survival:cox but that returns hazard ratios instead of time so I don't know how to separate them, it was surprisingly easy to add this one I just need to decide how to stop if the user tries to predict the wrong type given the objective he has chosen here is a reprex of it working
Created on 2022-03-15 by the reprex package (v2.0.1)