One of the most important hyper-parameters in the Random Forest (RF) algorithm is the feature set size used to search for the best partitioning rule at each node of trees. Most existing research on feature set size has been done primarily with a focus on classification problems. We studied the effect of feature set size in the context of regression. Through experimental studies using many datasets, we first investigated whether the RF regression predictions are affected by the feature set size. Then, we found a rule associated with the optimal size based on the characteristics of each data. Lastly, we developed a search algorithm for estimating the best feature set size in RF regression. We showed that the proposed search algorithm can provide improvements over other choices, such as using the default size specified in the randomForest R package and using the common grid search method.

Random forest is an ensemble method that combines many decision trees. Each level of trees is determined by an optimal rule among a candidate feature set. The candidate feature set is a random subset of all features, and is different at each level of trees. In this article, we investigated whether the accuracy of Random forest is affected by the size of the candidate feature set. We found that the optimal size differs from data to data without any specific pattern. To estimate the optimal size of feature set, we proposed a novel algorithm which uses the out-of-bag error and the ‘SearchSize’ exploration. The proposed method is significantly faster than the standard grid search method while giving almost the same accuracy. Finally, we demonstrated that the accuracy of Random forest using the proposed algorithm has increased significantly compared to using a typical size of feature set.

Background/Aims To investigate whether parapapillary choroidal microvasculature dropout (MvD) is associated with progressive retinal nerve fibre layer (RNFL) thinning in eyes with primary open-angle glaucoma (POAG) and disc haemorrhage (DH). Methods Parapapillary microvasculature was evaluated by swept-source optical coherence tomography (OCT) angiography (OCTA) in 50 eyes with POAG and DH, 1 year before, at the time of and 1 year after the detection of DH. MvD was defined as an area in the parapapillary deep layer of focal sectoral dropout with no visible microvascular network. Progressive changes in OCT RNFL thickness were compared in groups of eyes with and without MvD. Results Cumulative prevalence of MvD was 76.0% (38 eyes) at 1 year after detection of DH. All MvDs were detected in the same sectoral locations as DH. In eyes with MvD, global RNFL thickness and sectoral RNFL thickness at the location of DH were significantly reduced, both from 1 year before to the time of DH detection (both p<0.001) and from DH detection to 1 year later (both p<0.001). In eyes without MvD, however, the reductions in global (p=0.011) and sectoral (p=0.007) RNFL thickness were significant only from DH detection to 1 year later. Conclusion In eyes with POAG, RNFL thinning was spatially consistent and progressive at the location of MvD accompanied by subsequent DH and continued to progress after the occurrence of DH. When DH was not accompanied by MvD, progressive RNFL thinning was more likely to occur after the detection of DH.