xgb_to_code.R

library(xgboost)
library(reticulate)
#reticulate::py_install("m2cgen") # https://github.com/BayesWitnesses/m2cgen
#reticulate::py_install("xgboost")
#reticulate::py_install("sklearn")

# Fit an XGBoost Model (example from ?xgb.train)
data(agaricus.train, package='xgboost')
data(agaricus.test, package='xgboost')

dtrain <- with(agaricus.train, xgb.DMatrix(data, label = label, nthread = 2))
dtest <- with(agaricus.test, xgb.DMatrix(data, label = label, nthread = 2))
watchlist <- list(train = dtrain, eval = dtest)

param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2,
              objective = "binary:logistic", eval_metric = "auc")
bst <- xgb.train(param, dtrain, nrounds = 2, watchlist)
#> [14:51:40] WARNING: src/learner.cc:767: 
#> Parameters: { "verbose" } are not used.
#> 
#> [1]  train-auc:0.958228  eval-auc:0.960373 
#> [2]  train-auc:0.981413  eval-auc:0.979930

#Save to disk
xgboost::xgb.save(bst, "bst.xgb")
#> [1] TRUE

# Start a Python session
m2 <- reticulate::import("m2cgen")
xg <- reticulate::import("xgboost")

# Load the model from disk
mod <- xg$XGBRegressor()
mod$load_model(fname = "bst.xgb")

# Convert to a raw R function
eval(parse(text = m2$export_to_r(mod, function_name = "pred")))
pred
#> function(input) {
#>     if (input[29] >= -0.0000009536743) {
#>         if (input[109] >= -0.0000009536743) {
#>             var0 <- 1.8596492
#>         } else {
#>             var0 <- -1.9407086
#>         }
#>     } else {
#>         if (input[56] >= -0.0000009536743) {
#>             var0 <- -1.7004405
#>         } else {
#>             var0 <- 1.7121772
#>         }
#>     }
#>     if (input[60] >= -0.0000009536743) {
#>         var1 <- -6.2362447
#>     } else {
#>         if (input[29] >= -0.0000009536743) {
#>             var1 <- -0.96853036
#>         } else {
#>             var1 <- 0.78471756
#>         }
#>     }
#>     return(nan + (var0 + var1))
#> }
nan <- 0 # For some reason `nan` is added to the function output but it is undefined in R

# Predict with the original model
predict(bst,as.matrix(agaricus.test$data)[1,,drop = FALSE])
#> [1] 0.01241208

# Predict with the raw code
plogis(pred(as.matrix(agaricus.test$data[1, ])))
#> [1] 0.01241208

# Export as a Javascript function
cat(m2$export_to_javascript(mod, function_name = "pred"))
#> function pred(input) {
#>     var var0;
#>     if (input[28] >= -0.0000009536743) {
#>         if (input[108] >= -0.0000009536743) {
#>             var0 = 1.8596492;
#>         } else {
#>             var0 = -1.9407086;
#>         }
#>     } else {
#>         if (input[55] >= -0.0000009536743) {
#>             var0 = -1.7004405;
#>         } else {
#>             var0 = 1.7121772;
#>         }
#>     }
#>     var var1;
#>     if (input[59] >= -0.0000009536743) {
#>         var1 = -6.2362447;
#>     } else {
#>         if (input[28] >= -0.0000009536743) {
#>             var1 = -0.96853036;
#>         } else {
#>             var1 = 0.78471756;
#>         }
#>     }
#>     return nan + (var0 + var1);
#> }