How to have less bars on Bar graph?-CodePudding

I am building Random forest algorithm, the goal is to predict which features are more important. And I have Bar graph showing features importance from Random Forest Built-in Feature Importance. Is there a chance to filter out data that are relatively smaller compared to larger bars and how to implement this. I want to do these because there is a mess on these picture below:

Input code:

rf = RandomForestRegressor(n_estimators=100, max_depth=3)
rf.fit(X_train, y_train)

sorted_idx = rf.feature_importances_.argsort()
plt.figure(figsize=(8, 30))
plt.barh(X_train.columns[sorted_idx], rf.feature_importances_[sorted_idx])
plt.xlabel("Random Forest Feature Importance")

CodePudding user response：

By filtering the sorted_idx variable, you should be able to do so:

sorted_idx = rf.feature_importances_.argsort()[:5]

Instead of 5, you can obviously take whatever number of features you want to be plotted.