gblinear xgboost in R -
let's data has both numeric & catagoricial feature, , i've created xgboost model using gblinear. i've analyzed xgboost model xgb.importance, how can express categorical variable weights?
while xgboost considered black box model, can understand feature importance (for both categorical , numeric) averaging gain of each feature split , trees.
this represented in graph below.
# feature real names names <- dimnames(trainmatrix)[[2]] # compute feature importance matrix importance_matrix <- xgb.importance(names, model = bst) # nice graph xgb.plot.importance(importance_matrix[1:10,])
in feature importance above, can see first 10 important features.
this function gives color each bar. k-means clustering applied group each feature importance.
alternately, represented in tree diagram (see link above).
Comments
Post a Comment