xgboost : The meaning of the base_score parameter

parameters xgboost
rf7 picture rf7 · Dec 1, 2017 · Viewed 8.8k times · Source

In the documentation of xgboost I read:

base_score [default=0.5] : the initial prediction score of all instances, global bias

What is the meaning of this phrase? Is the base score the prior probability of the Event of Interest in the Dataset? I.e. in a dataset of 1,000 observations with 300 Positives and 700 Negatives the base score would be 0.3?

If not, what it would be?

Your advice will be appreciated.

Answer

Yue picture Yue · Mar 20, 2018

I think your understanding is correct, in your example the base score could be set to 0.3, or you can simply leave it to be the default 0.5. For highly imbalanced data you can initialize it to a more meaningful base score for an improved learning process. Theoretically, as long as you choose the right learning rate and give it enough steps to train, the starting base score shouldn't affect the result. Look at the author's answer in this issue.

Reference: https://github.com/dmlc/xgboost/issues/799

Related questions

How are parameters sent in an HTTP POST request?

In an HTTP GET request, parameters are sent as a query string: http://example.com/page?parameter=value&also;=another In an HTTP POST request, the parameters are not sent along with the URI. Where are the values? In …

Read More
What are the -Xms and -Xmx parameters when starting JVM?

Please explain the use of Xms and Xmx parameters in JVMs. What are the default values for them?

Read More
Does Java support default parameter values?

I came across some Java code that had the following structure: public MyParameterizedFunction(String param1, int param2) { this(param1, param2, false); } public MyParameterizedFunction(String param1, int param2, boolean param3) { //use all three parameters here } I know that in C++ I …

Read More