R Under development (unstable) (2026-10-05 r90641 ucrt) -- "Unsuffered Consequences" Copyright (C) 2026 The R Foundation for Statistical Computing Platform: x86_64-w64-mingw32/x64 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(testthat) > library(bartisan) > > test_check("bartisan") i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. Average treatment effect on the treated (difference) Treatment: `z` Averaged over the 62 units in group "high" contrast estimate lower upper n Y[high] - Y[low] 2.55 2.210 2.83 62 Y[high] - Y[mid] 1.21 0.888 1.52 62 Average potential outcomes quantity estimate lower upper Y[low] 0.579 0.384 0.804 Y[mid] 1.920 1.710 2.150 Y[high] 3.130 2.910 3.340 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". Average treatment effect on the treated (difference) Treatment: `z` Averaged over the 62 units in group "high" contrast estimate lower upper n Y[mid] - Y[low] 1.34 1.000 1.65 62 Y[high] - Y[low] 2.55 2.210 2.83 62 Y[high] - Y[mid] 1.21 0.888 1.52 62 Average potential outcomes quantity estimate lower upper Y[low] 0.579 0.384 0.804 Y[mid] 1.920 1.710 2.150 Y[high] 3.130 2.910 3.340 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". Average treatment effect on the treated (difference) Treatment: `z` Averaged over the 62 units in group "high" Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 1.42 1.16 1.69 200 Average potential outcomes quantity estimate lower upper Y[0] 0.568 0.38 0.763 Y[1] 1.990 1.84 2.160 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 1.42 1.16 1.69 200 Average potential outcomes quantity estimate lower upper Y[0] 0.568 0.38 0.763 Y[1] 1.990 1.84 2.160 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 1.42 1.12 1.63 200 Average potential outcomes quantity estimate lower upper Y[0] 0.568 0.37 0.736 Y[1] 1.990 1.83 2.150 i estimate is the posterior mean; lower and upper bound the 95% highest posterior density interval. i Y[a] is the average response with `z` set to "a". Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 1.42 1.12 1.63 200 Average potential outcomes quantity estimate lower upper Y[0] 0.568 0.37 0.736 Y[1] 1.990 1.83 2.150 i estimate is the posterior mean; lower and upper bound the 95% highest posterior density interval. i Y[a] is the average response with `z` set to "a". Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 1.65 1.45 1.86 200 Average potential outcomes quantity estimate lower upper Y[0] 0.321 0.139 0.491 Y[1] 1.970 1.810 2.120 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 1.65 1.45 1.86 200 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". Conditional average treatment effects (difference) Treatment: `z` contrast units min q25 median q75 max Y[1] - Y[0] 200 0.87 1.5 1.7 1.82 2 Average potential outcomes quantity estimate lower upper Y[0] 0.321 0.139 0.491 Y[1] 1.970 1.810 2.120 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the predicted response for that unit with `z` set to "a". i Quartiles of the per-unit estimates. The object itself holds one row per unit, with an interval each. Conditional average treatment effects (difference) Treatment: `z` contrast units min q25 median q75 max Y[1] - Y[0] 200 0.87 1.5 1.7 1.82 2 Average potential outcomes quantity estimate lower upper Y[0] 0.321 0.139 0.491 Y[1] 1.970 1.810 2.120 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the predicted response for that unit with `z` set to "a". i Quartiles of the per-unit estimates. The object itself holds one row per unit, with an interval each. Conditional average treatment effects (difference) Treatment: `z` contrast units min q25 median q75 max Y[1] - Y[0] 200 0.87 1.5 1.7 1.82 2 Average potential outcomes quantity estimate lower upper Y[0] 0.321 0.139 0.491 Y[1] 1.970 1.810 2.120 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the predicted response for that unit with `z` set to "a". i Quartiles of the per-unit estimates. The object itself holds one row per unit, with an interval each. Nuisance parameters mean sd lower upper sigma 0.918 0.049 0.827 1.011 b.z.0 -0.565 0.132 -0.859 -0.307 b.z.1 0.222 0.306 -0.059 1.019 Convergence and mixing Log likelihood: R-hat 1.013, bulk ESS 81, tail ESS 111, over 1 chain i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors in the "(Intercept)" forest ranked by use; 4 of 4 shown. variable prop_used prop_splits splits x1 1 0.339 5.3 .propensity 1 0.249 4.0 g 1 0.209 3.3 x2 1 0.202 3.3 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. i This fit has a treatment, "z". estimate_effect() (`?bartisan::estimate_effect`) reports its effect, with the average potential outcomes beside it. Nuisance parameters mean sd lower upper sigma 0.918 0.049 0.827 1.011 b.z.0 -0.565 0.132 -0.859 -0.307 b.z.1 0.222 0.306 -0.059 1.019 Convergence and mixing Log likelihood: R-hat 1.013, bulk ESS 81, tail ESS 111, over 1 chain i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors in the "(Intercept)" forest ranked by use; 4 of 4 shown. variable prop_used prop_splits splits x1 1 0.339 5.3 .propensity 1 0.249 4.0 g 1 0.209 3.3 x2 1 0.202 3.3 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. i This fit has a treatment, "z". estimate_effect() (`?bartisan::estimate_effect`) reports its effect, with the average potential outcomes beside it. Nuisance parameters mean sd lower upper sigma 1.222 0.062 1.122 1.342 Convergence and mixing Log likelihood: R-hat 1.113, bulk ESS 20, tail ESS 40, over 1 chain i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors ranked by use; 2 of 2 shown. variable prop_used prop_splits splits x2 1 0.593 4.7 x1 1 0.407 3.2 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Generalized BART Call: bcf(formula = y ~ x1 + x2 + g, treat = ~z, data = d, family = if (binary) stats::binomial() else stats::gaussian(), control = quick_control(num_trees = 10L, num_burn = 50L, num_draws = 100L)) Family: "gaussian" with the "identity" link Observations: 200 Structure: 2 forests of 10 trees, soft decision rules Draws: 100 kept after 50 warmup Posterior means: sigma = 0.918, b.z.0 = -0.565, b.z.1 = 0.222 Treatment: "z" Effect moderators: "x1", "x2", and "g" i `estimate_effect()` reports the treatment effect, with the average potential outcomes beside it; `plot()` draws the conditional ones. Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 0.232 0.101 0.364 200 Average potential outcomes quantity estimate lower upper Y[0] 0.495 0.409 0.593 Y[1] 0.727 0.643 0.818 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 0.232 0.101 0.364 200 Average potential outcomes quantity estimate lower upper Y[0] 0.495 0.409 0.593 Y[1] 0.727 0.643 0.818 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". Average treatment effect (log odds ratio) Treatment: `z` Averaged over 200 units contrast estimate lower upper n log(O(Y[1]) / O(Y[0])) 1.01 0.421 1.62 200 Average potential outcomes quantity estimate lower upper Y[0] 0.495 0.409 0.593 Y[1] 0.727 0.643 0.818 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a", and O(y) is the odds `y/(1-y)`. Average treatment effect (log odds ratio) Treatment: `z` Averaged over 200 units contrast estimate lower upper n log(O(Y[1]) / O(Y[0])) 1.01 0.421 1.62 200 Average potential outcomes quantity estimate lower upper Y[0] 0.495 0.409 0.593 Y[1] 0.727 0.643 0.818 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a", and O(y) is the odds `y/(1-y)`. Average treatment effect (difference) Treatment: `z` Averaged over 200 units contrast estimate lower upper n Y[1] - Y[0] 0.232 0.101 0.364 200 Average potential outcomes quantity estimate lower upper Y[0] 0.495 0.409 0.593 Y[1] 0.727 0.643 0.818 i estimate is the posterior mean; lower and upper bound the 95% equal-tailed credible interval. i Y[a] is the average response with `z` set to "a". i The "half-probit" link's inverse does not cover the whole additive predictor, so some proposals will have a non-finite density and be rejected. i The fit is valid but slower and less accurate; a link whose inverse is defined on the whole line, such as "logit", avoids it. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. Variable importance variable prop_used prop_splits splits x1 1.000 0.951 14.6 x2 0.300 0.047 0.7 x3 0.017 0.001 0.0 i splits_lower and splits_upper hold the 95% interval, not shown above. Variable importance variable prop_used prop_splits splits x1 1.000 0.951 14.6 x2 0.300 0.047 0.7 x3 0.017 0.001 0.0 i splits_lower and splits_upper hold the 95% interval, not shown above. Variable importance variable prop_used prop_splits splits x1 1.000 0.951 14.6 x2 0.300 0.047 0.7 x3 0.017 0.001 0.0 i splits_lower and splits_upper hold the 95% interval, not shown above. Variable importance variable prop_used prop_splits splits x1 1.000 0.951 14.6 x2 0.300 0.047 0.7 x3 0.017 0.001 0.0 i splits_lower and splits_upper hold the 95% interval, not shown above. Variable importance variable prop_used prop_splits splits x1 1.000 0.410 6.6 x2 1.000 0.277 4.3 x3 0.933 0.313 5.4 i Fitted with `sparsity = FALSE` (the default), so every predictor keeps a share of the rules and prop_used is near 1 throughout. Refit with `sparsity = TRUE` to read it as a selection rule. i splits_lower and splits_upper hold the 95% interval, not shown above. Variable importance variable prop_used prop_splits splits x1 1.000 0.410 6.6 x2 1.000 0.277 4.3 x3 0.933 0.313 5.4 i Fitted with `sparsity = FALSE` (the default), so every predictor keeps a share of the rules and prop_used is near 1 throughout. Refit with `sparsity = TRUE` to read it as a selection rule. i splits_lower and splits_upper hold the 95% interval, not shown above. Generalized BART Call: bartisan(formula = y ~ x1 + x2 + (1 | g), data = d, family = stats::gaussian(), control = ctrl) Family: "gaussian" with the "identity" link Observations: 100 Structure: 1 forest of 5 trees, soft decision rules Draws: 10 kept after 10 warmup Random intercepts: g (4 levels) Posterior means: sigma = 1.06 Generalized BART Call: bartisan(formula = y ~ x1 + x2 + (1 | g), data = d, family = stats::gaussian(), control = ctrl) Family: "gaussian" with the "identity" link Observations: 100 Structure: 1 forest of 5 trees, soft decision rules Draws: 10 kept after 10 warmup Random intercepts: g (4 levels) Posterior means: sigma = 1.06 Nuisance parameters mean sd lower upper sigma 1.056 0.056 1.002 1.17 Random-effect scales Standard deviation of the group intercepts. mean sd lower upper g 0.573 0.092 0.416 0.726 Convergence and mixing Log likelihood: R-hat 0.962, bulk ESS NA, tail ESS NA, over 1 chain i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors ranked by use; 2 of 2 shown. variable prop_used prop_splits splits x2 1 0.571 3.4 x1 1 0.429 2.7 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Nuisance parameters mean sd lower upper sigma 1.056 0.056 1.002 1.17 Random-effect scales Standard deviation of the group intercepts. mean sd lower upper g 0.573 0.092 0.416 0.726 Convergence and mixing Log likelihood: R-hat 0.962, bulk ESS NA, tail ESS NA, over 1 chain i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors ranked by use; 2 of 2 shown. variable prop_used prop_splits splits x2 1 0.571 3.4 x1 1 0.429 2.7 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Generalized BART Call: bartisan(formula = y ~ x1 + x2 + (1 | g), data = d, family = gaussian_ls(), control = quick_control(num_trees = c(mean = 5L, log_sd = 3L), num_burn = 10L, num_draws = 10L)) Family: "gaussian_ls" with the "identity" link Observations: 100 Structure: 2 forests of 5 and 3 trees, soft decision rules Draws: 10 kept after 10 warmup Random intercepts: g (4 levels) Generalized BART Call: bartisan(formula = y ~ x1 + x2 + (1 | g), data = d, family = gaussian_ls(), control = quick_control(num_trees = c(mean = 5L, log_sd = 3L), num_burn = 10L, num_draws = 10L)) Family: "gaussian_ls" with the "identity" link Observations: 100 Structure: 2 forests of 5 and 3 trees, soft decision rules Draws: 10 kept after 10 warmup Random intercepts: g (4 levels) Generalized BART Call: bartisan(formula = b ~ x1 + x2, data = d, family = stats::binomial("cauchit"), control = ctrl) Family: "binomial" with the "cauchit" link (supplied from R) Observations: 90 Structure: 1 forest of 5 trees, soft decision rules Draws: 10 kept after 10 warmup Generalized BART Call: bartisan(formula = b ~ x1 + x2, data = d, family = stats::binomial("cauchit"), control = ctrl) Family: "binomial" with the "cauchit" link (supplied from R) Observations: 90 Structure: 1 forest of 5 trees, soft decision rules Draws: 10 kept after 10 warmup Generalized BART Call: bartisan(formula = m ~ x1 + x2, data = d, family = multinomial("probit"), control = ctrl) Family: "multinomial" with the "probit" link Observations: 90 Structure: 2 forests of 5 trees, soft decision rules Draws: 10 kept after 10 warmup Posterior means: sigma11 = 0.459, sigma21 = 0.11, sigma22 = 1.54 Generalized BART Call: bartisan(formula = m ~ x1 + x2, data = d, family = multinomial("probit"), control = ctrl) Family: "multinomial" with the "probit" link Observations: 90 Structure: 2 forests of 5 trees, soft decision rules Draws: 10 kept after 10 warmup Posterior means: sigma11 = 0.459, sigma21 = 0.11, sigma22 = 1.54 Generalized BART Call: bartisan(formula = y ~ x1 + x2, data = d, family = normal, control = ctrl) Family: "unit normal" (supplied from R) Observations: 90 Structure: 1 forest of 5 trees, soft decision rules Draws: 10 kept after 10 warmup Generalized BART Call: bartisan(formula = y ~ x1 + x2, data = d, family = normal, control = ctrl) Family: "unit normal" (supplied from R) Observations: 90 Structure: 1 forest of 5 trees, soft decision rules Draws: 10 kept after 10 warmup Nuisance parameters mean sd lower upper sigma 1.123 0.148 0.92 1.342 Convergence and mixing i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors ranked by use; 2 of 2 shown. variable prop_used prop_splits splits x1 1.0 0.758 5.0 x2 0.9 0.242 1.7 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Nuisance parameters mean sd lower upper sigma 1.123 0.148 0.92 1.342 Convergence and mixing i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors ranked by use; 2 of 2 shown. variable prop_used prop_splits splits x1 1.0 0.758 5.0 x2 0.9 0.242 1.7 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Nuisance parameters mean sd lower upper sigma 1.123 0.148 0.92 1.342 Convergence and mixing i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors ranked by use; 2 of 2 shown. variable prop_used prop_splits splits x1 1.0 0.758 5.0 x2 0.9 0.242 1.7 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Nuisance parameters mean sd lower upper sigma 1.123 0.148 0.92 1.342 Convergence and mixing i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors ranked by use; 2 of 2 shown. variable prop_used prop_splits splits x1 1.0 0.758 5.0 x2 0.9 0.242 1.7 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Nuisance parameters mean sd lower upper sigma 1.123 0.148 0.92 1.342 Convergence and mixing Log likelihood: R-hat 1.402, bulk ESS NA, tail ESS NA, over 1 chain i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Nuisance parameters mean sd lower upper sigma 1.123 0.148 0.92 1.342 Convergence and mixing Log likelihood: R-hat 1.402, bulk ESS NA, tail ESS NA, over 1 chain i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. Nuisance parameters mean sd lower upper lambda1 0.394 0.139 0.206 0.620 lambda2 0.309 0.072 0.216 0.421 lambda3 0.374 0.157 0.158 0.640 lambda4 0.563 0.148 0.347 0.786 lambda5 0.415 0.166 0.156 0.639 lambda6 0.307 0.079 0.214 0.451 lambda8 0.730 0.404 0.403 1.402 lambda9 0.397 0.173 0.193 0.696 lambda10 0.239 0.102 0.090 0.405 lambda11 0.660 0.280 0.337 1.089 lambda12 0.408 0.170 0.111 0.637 lambda_rate 2.848 0.877 1.795 4.130 1 more, omitted; all of them are in `fit$aux`. Convergence and mixing Log likelihood: R-hat 1.11, bulk ESS NA, tail ESS NA, over 1 chain i Use diagnose() (`?bartisan::diagnose`) to examine convergence and mixing diagnostics. Variable importance Predictors ranked by use; 2 of 2 shown. variable prop_used prop_splits splits x1 1.0 0.591 3.3 x2 0.9 0.409 2.6 i Use variable_importance() (`?bartisan::variable_importance`) to examine variable importance. Further tools i Use loo() (`?bartisan::loo.bartisan_fit`) to compare this fit with others, or kfold() (`?bartisan::kfold.bartisan_fit`) if `loo()` reports many Pareto k values above 0.7. i Use partial_dependence() (`?bartisan::partial_dependence`) and plot() (`?bartisan::plot.bartisan_fit`) to view the partial dependence of the predictions on a predictor. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. Priors forest num_trees gamma leaf_scale mean 5 0.95 0.691 log_sd 3 0.50 0.433 Trees * num_trees trees per additive predictor, summed. A node at depth d branches with probability gamma * (1 + d)^-2. Leaves * Each leaf value is Normal(0, leaf_scale^2), that scale being 3 * s / (2 * sqrt(num_trees)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated for mean and held fixed for the rest. Splitting variables * Each of the 2 predictors is equally likely to be split on, and that is not drawn (`sparsity = FALSE`). Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is held fixed. Family: gaussian_ls, identity link * No parameters of its own beyond the additive predictors above. i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.677^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * The share of the rules each of the 2 predictors receives is Dirichlet(1 / 2), whose concentration enters as a / (a + 2) ~ Beta(0.5, 1). Both are estimated. Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Family: gaussian, identity link * sigma: HalfCauchy(0, 0.9617) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.677^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * The share of the rules each of the 2 predictors receives is Dirichlet(1 / 2), whose concentration enters as a / (a + 2) ~ Beta(0.5, 1). The shares are estimated and the concentration is held fixed. Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Family: gaussian, identity link * sigma: HalfCauchy(0, 0.9617) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.677^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * The share of the rules each of the 2 predictors receives is Dirichlet(1 / 2), whose concentration enters as a / (a + 2) ~ Beta(0.5, 1). The shares are estimated and the concentration is held fixed. Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Family: gaussian, identity link * sigma: HalfCauchy(0, 0.9617) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.677^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * The share of the rules each of the 2 predictors receives is Dirichlet(1 / 2), whose concentration enters as a / (a + 2) ~ Beta(0.5, 1). Both are held fixed. Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Family: gaussian, identity link * sigma: HalfCauchy(0, 0.9617) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.677^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * The share of the rules each of the 2 predictors receives is Dirichlet(1 / 2), whose concentration enters as a / (a + 2) ~ Beta(0.5, 1). Both are held fixed. Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Family: gaussian, identity link * sigma: HalfCauchy(0, 0.9617) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors forest leaf_scale mean 0.677 log_sd 0.335 Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, leaf_scale^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * The share of the rules each of the 2 predictors receives is Dirichlet(1 / 2), whose concentration enters as a / (a + 2) ~ Beta(0.5, 1). Both are estimated. * The forests share one set of shares (`share_sparsity = TRUE`). Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Family: gaussian_ls, identity link * No parameters of its own beyond the additive predictors above. i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors forest leaf_scale mean 0.677 log_sd 0.335 Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, leaf_scale^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * The share of the rules each of the 2 predictors receives is Dirichlet(1 / 2), whose concentration enters as a / (a + 2) ~ Beta(0.5, 1). Both are estimated. * The forests share one set of shares (`share_sparsity = TRUE`). Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Family: gaussian_ls, identity link * No parameters of its own beyond the additive predictors above. i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.785^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * Each of the 1 predictors is equally likely to be split on, and that is not drawn (`sparsity = FALSE`). Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Group intercepts * Term "g", with 4 levels. Each level's intercept is Normal(0, tau^2), and tau is given the half-Cauchy prior the leaf scale gets, centred at 0.785; it is drawn. Family: gaussian, identity link * sigma: HalfCauchy(0, 1.155) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.785^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * Each of the 1 predictors is equally likely to be split on, and that is not drawn (`sparsity = FALSE`). Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Group intercepts * Term "g", with 4 levels. Each level's intercept is Normal(0, tau^2), and tau is given the half-Cauchy prior the leaf scale gets, centred at 0.785; it is drawn. Family: gaussian, identity link * sigma: HalfCauchy(0, 1.155) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.785^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * Each of the 1 predictors is equally likely to be split on, and that is not drawn (`sparsity = FALSE`). Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Group intercepts * Term "g", with 4 levels. Each level's intercept is Normal(0, tau^2), and tau is given the half-Cauchy prior the leaf scale gets, centred at 0.785; it is held at that prior's median. Family: gaussian, identity link * sigma: HalfCauchy(0, 1.155) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. Priors Trees * 5 trees per additive predictor, summed. A node at depth d branches with probability 0.95 * (1 + d)^-2, so the root splits with probability 0.95 and a node at depth 3 with 0.059. Leaves * Each leaf value is Normal(0, 0.785^2), that scale being 3 * s / (2 * sqrt(5)) with s the response's scale on the link scale. The scale is itself given a half-Cauchy prior centred there and is estimated. Splitting variables * Each of the 1 predictors is equally likely to be split on, and that is not drawn (`sparsity = FALSE`). Decision rules * Soft, with "smoothstep" gates. Each tree's bandwidth is drawn from an exponential with mean 0.1, on predictors mapped to [0, 1], and is estimated. Group intercepts * Term "g", with 4 levels. Each level's intercept is Normal(0, tau^2), and tau is given the half-Cauchy prior the leaf scale gets, centred at 0.785; it is held at that prior's median. Family: gaussian, identity link * sigma: HalfCauchy(0, 1.155) (estimated) i The leaf scale, and any number above read off the response, are calibrated rather than fitted; that is how a BART prior is specified. i `prior_only = TRUE` in `bartisan()` draws from all of this, so that what it implies can be read on the outcome's own scale. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. i Using `family = dpm()`. i Set `family` explicitly to silence this message. [ FAIL 0 | WARN 0 | SKIP 167 | PASS 2036 ] ══ Skipped tests (167) ═════════════════════════════════════════════════════════ • On CRAN (166): 'test-augment.R:7:3', 'test-augment.R:90:3', 'test-augment.R:120:3', 'test-augment.R:261:3', 'test-augment.R:299:3', 'test-augment.R:337:3', 'test-augment.R:410:3', 'test-augment.R:437:3', 'test-augment.R:465:3', 'test-augment.R:527:3', 'test-augment.R:596:3', 'test-bartisan.R:515:3', 'test-bcf.R:105:3', 'test-bcf.R:173:3', 'test-bcf.R:302:3', 'test-bcf.R:363:3', 'test-bookkeeping.R:32:3', 'test-categorical.R:16:3', 'test-categorical.R:37:3', 'test-chains.R:175:3', 'test-chains.R:239:3', 'test-chains.R:266:3', 'test-chains.R:291:3', 'test-chains.R:349:3', 'test-chains.R:358:3', 'test-chains.R:412:3', 'test-chains.R:449:3', 'test-chains.R:533:3', 'test-control.R:171:3', 'test-control.R:204:3', 'test-density.R:165:3', 'test-density.R:193:3', 'test-diagnose.R:215:3', 'test-diagnose.R:280:3', 'test-dpm.R:41:3', 'test-dpm.R:86:3', 'test-dpm.R:130:3', 'test-dpm.R:176:3', 'test-dpm.R:242:3', 'test-dpm.R:259:3', 'test-dpm.R:324:3', 'test-dpm.R:376:3', 'test-estimate-effect.R:98:3', 'test-estimate-effect.R:503:3', 'test-estimate-effect.R:595:3', 'test-estimate-effect.R:708:3', 'test-exponential.R:46:3', 'test-exponential.R:289:3', 'test-families.R:308:3', 'test-families.R:343:3', 'test-flexible.R:368:3', 'test-flexible.R:443:3', 'test-flexible.R:511:3', 'test-flexible.R:563:3', 'test-flexible.R:843:3', 'test-gate.R:31:3', 'test-gate.R:69:3', 'test-interop.R:25:3', 'test-interop.R:91:3', 'test-interop.R:126:3', 'test-interop.R:171:3', 'test-interop.R:270:3', 'test-interop.R:588:3', 'test-interop.R:831:3', 'test-interop.R:889:3', 'test-interop.R:916:3', 'test-interop.R:953:3', 'test-interop.R:1022:3', 'test-interop.R:1085:3', 'test-interop.R:1137:3', 'test-interop.R:1217:3', 'test-interop.R:1406:3', 'test-invariants.R:53:3', 'test-invariants.R:73:3', 'test-invariants.R:102:3', 'test-invariants.R:184:3', 'test-invariants.R:399:3', 'test-invariants.R:607:3', 'test-invariants.R:716:3', 'test-invariants.R:793:3', 'test-invariants.R:819:3', 'test-invariants.R:845:3', 'test-invariants.R:862:3', 'test-invariants.R:883:3', 'test-latent.R:82:3', 'test-latent.R:121:3', 'test-latent.R:181:3', 'test-latent.R:230:3', 'test-latent.R:268:3', 'test-marginaleffects.R:109:3', 'test-marginaleffects.R:179:3', 'test-marginaleffects.R:245:3', 'test-marginaleffects.R:288:3', 'test-marginaleffects.R:373:3', 'test-methods.R:2:3', 'test-methods.R:69:3', 'test-missing.R:146:3', 'test-missing.R:175:3', 'test-mnp.R:35:3', 'test-mnp.R:75:3', 'test-mnp.R:103:3', 'test-mnp.R:147:3', 'test-newfamilies.R:35:3', 'test-newfamilies.R:133:3', 'test-newfamilies.R:165:3', 'test-ordinal.R:6:3', 'test-ordinal.R:43:3', 'test-ordinal.R:72:3', 'test-ordinal.R:162:3', 'test-ordinal.R:221:3', 'test-partial-dependence.R:168:3', 'test-perforest.R:17:3', 'test-perforest.R:94:3', 'test-perforest.R:267:3', 'test-predict.R:169:3', 'test-prior-only.R:7:3', 'test-prior-only.R:71:3', 'test-prior-only.R:103:3', 'test-prior-only.R:157:3', 'test-progress.R:142:3', 'test-quadratic.R:82:3', 'test-random.R:7:3', 'test-random.R:48:3', 'test-random.R:141:3', 'test-random.R:178:3', 'test-random.R:247:3', 'test-ranef.R:53:3', 'test-recovery.R:5:3', 'test-recovery.R:46:3', 'test-recovery.R:76:3', 'test-recovery.R:106:3', 'test-recovery.R:122:3', 'test-recovery.R:150:3', 'test-recovery.R:181:3', 'test-recovery.R:259:3', 'test-recovery.R:301:3', 'test-recovery.R:322:3', 'test-recovery.R:408:3', 'test-recovery.R:428:3', 'test-recovery.R:503:3', 'test-recovery.R:548:3', 'test-recovery.R:607:3', 'test-share-forests.R:100:3', 'test-share-forests.R:126:3', 'test-share-forests.R:142:3', 'test-share-forests.R:166:3', 'test-share-forests.R:203:3', 'test-share-forests.R:221:3', 'test-share-sparsity.R:61:3', 'test-share-sparsity.R:96:3', 'test-share-sparsity.R:121:3', 'test-tweedie.R:241:3', 'test-tweedie.R:278:3', 'test-tweedie.R:300:3', 'test-tweedie.R:324:3', 'test-varying.R:287:3', 'test-varying.R:353:3', 'test-varying.R:398:3', 'test-varying.R:462:3', 'test-varying.R:523:3', 'test-varying.R:577:3', 'test-varying.R:604:3', 'test-varying.R:696:3', 'test-varying.R:837:3', 'test-varying.R:986:3', 'test-varying.R:1068:3' • this chain did split on x1, so there is nothing to check (1): 'test-partial-dependence.R:362:3' [ FAIL 0 | WARN 0 | SKIP 167 | PASS 2036 ] > > proc.time() user system elapsed 203.65 11.29 215.43