Error with single_prediction function: Error in UseMethod("broken")

[kransom14]

When I try to run the vignette examples for the single_prediction() function I see the following error for the random forest model:

Error in UseMethod("broken") :
no applicable method for 'broken' applied to an object of class "c('randomForest.formula', 'randomForest')"

I have DALEX package version 0.1 and breakDown 0.1.3

[pbiecek]

Try breakDown 1.4 from github
https://github.com/pbiecek/breakDown
Version 1.3 supports only lm and glm model while 1.4 is model agnostic

[kransom14]

Great, thank you. I now have breakDown 1.4 and get this error when I run single_prediction() with an explainer created for a gbm model: Error in paste("Using", n.trees, "trees...\n") : argument "n.trees" is missing, with no default The same error comes up if I set n.trees inside the single_prediction function.

…

On Tue, Feb 20, 2018 at 5:16 PM, Przemysław Biecek ***@***.*** > wrote: Try breakDown 1.4 from github https://github.com/pbiecek/breakDown Version 1.3 supports only lm and glm model while 1.4 is model agnostic — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#4 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ARUTZj36WBo_GmXu8aCWDo0t52Ww_Ckqks5tW25pgaJpZM4SMe8o> .

-- -- Katherine Ransom, PhD Hydrologic Sciences, UC Davis

[pbiecek]

Would you provide the full example for gbm? Will be easier to debug

[pbiecek]

Ok, I guess that your new observation is neither a data.frame nor matrix?

[kransom14]

Hi, The new observation is a data.frame. Below I made an example using the wine data. library(gbm) library(DALEX) library(breakDown) # create a gbm model model <- gbm(quality ~ pH + residual.sugar + sulphates + alcohol, data = wine, distribution = "gaussian", n.trees = 1000, interaction.depth = 4, shrinkage = 0.01, n.minobsinnode = 10, verbose = FALSE) # make an explainer for the model explainer_gbm <- explain(model, data = wine) # create a new observation new.wine <- data.frame(citric.acid = 0.35, sulphates = 0.6, alcohol = 12.5, pH = 3.36, residual.sugar = 4.8) # single prediction explained, this gives the n.trees error exp_sgn <- single_prediction(explainer_gbm, observation = new.wine) #Error in paste("Using", n.trees, "trees...\n") : # argument "n.trees" is missing, with no default # try adding n.trees as an argument, gives same error exp_sgn <- single_prediction(explainer_gbm, observation = new.wine, n.trees = 1000) #Error in paste("Using", n.trees, "trees...\n") : # argument "n.trees" is missing, with no default Session info: sessionInfo() R version 3.4.3 (2017-11-30) Platform: x86_64-w64-mingw32/x64 (64-bit) Running under: Windows 7 x64 (build 7601) Service Pack 1 Matrix products: default locale: [1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United States.1252 [3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C [5] LC_TIME=English_United States.1252 attached base packages: [1] parallel splines stats graphics grDevices utils datasets methods base other attached packages: [1] breakDown_0.1.4 DALEX_0.1 gbm_2.1.3 lattice_0.20-35 survival_2.41-3 devtools_1.13.4 loaded via a namespace (and not attached): [1] Rcpp_0.12.14 magrittr_1.5 munsell_0.4.3 colorspace_1.3-2 R6_2.2.2 rlang_0.1.6 [7] httr_1.3.1 plyr_1.8.4 tools_3.4.3 ALEPlot_1.0 grid_3.4.3 yaImpute_1.0-29 [13] gtable_0.2.0 git2r_0.21.0 withr_2.1.1 yaml_2.1.16 lazyeval_0.2.1 digest_0.6.14 [19] tibble_1.4.1 pdp_0.6.0 Matrix_1.2-12 gridExtra_2.3 ggplot2_2.2.1 curl_3.1 [25] memoise_1.1.0 pillar_1.1.0 compiler_3.4.3 scales_0.5.0 On Tue, Feb 20, 2018 at 9:32 PM, Przemysław Biecek <[email protected]> wrote: Ok, I guess that your new observation is neither a data.frame nor matrix?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#4 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ARUTZlJxvhIwy_GCYIF3gAMEfCQHnYwdks5tW6qKgaJpZM4SMe8o> .

​ -- -- Katherine Ransom, PhD Hydrologic Sciences, UC Davis

[pbiecek]

n.trees argument is required by predict.gbm but was not correctly passed from DALEX to breakDown.
Please update both packages and following example is working for me

library(gbm)
library(DALEX)
library(breakDown)

# create a gbm model
model <- gbm(quality ~ pH + residual.sugar + sulphates + alcohol, data = wine,
             distribution = "gaussian",
             n.trees = 1000,
             interaction.depth = 4,
             shrinkage = 0.01,
             n.minobsinnode = 10,
             verbose = FALSE)

# make an explainer for the model
explainer_gbm <- explain(model, data = wine)

# create a new observation
new.wine <- data.frame(citric.acid = 0.35,
                       sulphates = 0.6,
                       alcohol = 12.5,
                       pH = 3.36,
                       residual.sugar = 4.8)

exp_sgn <- single_prediction(explainer_gbm, observation = new.wine, n.trees = 1000)

exp_sgn
plot(exp_sgn)

[kransom14]

This fixes it, thank you. I will close the issue.

[thormic]

If anyone still had troubles with the first error (even after updating breakDown) - check if you are using explain from DALEX or from dplyr. If the latter is the case, just use DALEX::explain.

[bkochanski]

If anyone still had troubles with the first error (even after updating breakDown) - check if you are using explain from DALEX or from dplyr. If the latter is the case, just use DALEX::explain.

Thanks! Now I know what caused Błąd w poleceniu 'UseMethod("explain")': niestosowalna metoda dla 'explain' zastosowana do obiektu klasy "c('glm', 'lm')" ;-)