Function to visualize the XGBoost explainer in Python.
The XGBoost explainer in Python is cloned from: https://github.com/gameofdimension/xgboost_explainer and combined with my plotting code.
The function will plot the contribution of each variable to the final prediction in the order of absolute contribution.
# import
import xgboost as xgb
from xgboost_visualizer import *
# train a XGBoost model using xgb.train() rather than the sklearn equivalent interface.
xgb_trained = xgb.train(dtrain, ...)
# get a feature list for the variables used in the model training
features = [...]
# get a sample
sample = xgb.DMatrix(...)
# plot contribution
plot_contribution(xgb_trained, sample, features)The output plot will look like the following:
