Jackknife variance estimates for random forest

In statistics, jackknife variance estimates for random forest are a way to estimate the variance in random forest models, in order to eliminate the bootstrap effects.

Jackknife variance estimates

edit

The sampling variance of bagged learners is:

 

Jackknife estimates can be considered to eliminate the bootstrap effects. The jackknife variance estimator is defined as:[1]

 

In some classification problems, when random forest is used to fit models, jackknife estimated variance is defined as:

 

Here,  denotes a decision tree after training,   denotes the result based on samples without   observation.

Examples

edit

E-mail spam problem is a common classification problem, in this problem, 57 features are used to classify spam e-mail and non-spam e-mail. Applying IJ-U variance formula to evaluate the accuracy of models with m=15,19 and 57. The results shows in paper( Confidence Intervals for Random Forests: The jackknife and the Infinitesimal Jackknife ) that m = 57 random forest appears to be quite unstable, while predictions made by m=5 random forest appear to be quite stable, this results is corresponding to the evaluation made by error percentage, in which the accuracy of model with m=5 is high and m=57 is low.

Here, accuracy is measured by error rate, which is defined as:

 

Here N is also the number of samples, M is the number of classes,   is the indicator function which equals 1 when   observation is in class j, equals 0 when in other classes. No probability is considered here. There is another method which is similar to error rate to measure accuracy:

 

Here N is the number of samples, M is the number of classes,   is the indicator function which equals 1 when   observation is in class j, equals 0 when in other classes.   is the predicted probability of   observation in class  .This method is used in Kaggle[2] These two methods are very similar.

Modification for bias

edit

When using Monte Carlo MSEs for estimating   and  , a problem about the Monte Carlo bias should be considered, especially when n is large, the bias is getting large:

 

To eliminate this influence, bias-corrected modifications are suggested:

 
 

References

edit
  1. ^ Wager, Stefan; Hastie, Trevor; Efron, Bradley (2014-05-14). "Confidence Intervals for Random Forests: The Jackknife and the Infinitesimal Jackknife". Journal of Machine Learning Research. 15 (1): 1625–1651. arXiv:1311.4555. Bibcode:2013arXiv1311.4555W. PMC 4286302. PMID 25580094.
  2. ^ "Otto Group Product Classification Challenge". Kaggle.