1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
|
##### Version 0.17.0
* Set num.threads=2 as default; respect environment variables and options
* Add handling of missing values for classification and regression
* Allow vector min.node.size and min.bucket for class-specific limits
* Add Poisson splitting rule for regression trees
* Add hierarchical shrinkage
* Fix a bug for always.split.variables (for some settings)
##### Version 0.16.0
* Add node.stats option to save node statistics of all nodes
* Add time.interest option to restrict unique survival times (faster and saves memory)
* Fix min bucket option in C++ version
* Fix memory error for always.split.variables in certain settings
* Fix quantile regression for factor variables in "order" mode
##### Version 0.15.0
* Switch to C++14 standard
* Add min.bucket parameter to restrict terminal node size
* Fix a bug with always.split.variables selecting the wrong variables
##### Version 0.14.0
* Faster permutation variable importance for high dimensional data (thanks to Roman Hornung)
* Add deforest() function to remove trees from ensemble
* Allow split.select.weights and always.split.variables together
* Add as.data.frame() method for predictions
* Fix weight calculation in case-specific RF (csrf())
* Fix cross compiling for Windows
##### Version 0.13.0
* Faster quantile prediction
* Add ... argument to ranger()
* Bug fixes
##### Version 0.12.0
* Faster computation (in some cases)
* Add local variable importance
* Add "hellinger" splitrule for binary classification
* Add "beta" splitrule for bounded outcomes
* Accept user-specified function in quantile prediction
* Add regularization
* Add x/y interface
* Internal changes (seed differences possible, prediction incompatible with older versions)
* Bug fixes
##### Version 0.11.0
* Add max.depth parameter to limit tree depth
* Add inbag argument for manual selection of observations in trees
* Add support of splitting weights for corrected impurity importance
* Internal changes (slightly improved computation speed)
* Warning: Possible seed differences compared to older versions
* Bug fixes
##### Version 0.10.0
* Change license of C++ core to MIT (R package is still GPL3)
* Better 'order' mode for unordered factors for multiclass and survival
* Add 'order' mode for unordered factors for GenABEL SNP data (binary classification and regression)
* Add class-weighted Gini splitting
* Add fixed proportion sampling
* Add impurity importance for the maxstat splitting rule
* Remove GenABEL from suggested packages (removed from CRAN). GenABEL data is still supported
* Improve memory management (internal changes)
* Bug fixes
##### Version 0.9.0
* Add bias-corrected impurity importance (actual impurity reduction, AIR)
* Add quantile prediction as in quantile regression forests
* Add treeInfo() function to extract human readable tree structure
* Add standard error estimation with the infinitesimal jackknife (now the default)
* Add impurity importance for survival forests
* Faster aggregation of predictions
* Fix memory issues on Windows 7
* Bug fixes
##### Version 0.8.0
* Handle sparse data of class Matrix::dgCMatrix
* Add prediction of standard errors to predict()
* Allow devtools::install_github() without subdir and on Windows
* Bug fixes
##### Version 0.7.0
* Add randomized splitting (extraTrees)
* Better formula interface: Support interactions terms and faster computation
* Split at mid-point between candidate values
* Improvements in holdoutRF and importance p-value estimation
* Drop unused factor levels in outcome before growing
* Add predict.all for probability and survival prediction
* Bug fixes
##### Version 0.6.0
* Set write.forest=TRUE by default
* Add num.trees option to predict()
* Faster version of getTerminalNodeIDs(), included in predict()
* Handle new factor levels in 'order' mode
* Use unadjusted p-value for 2 categories in maxstat splitting
* Bug fixes
##### Version 0.5.0
* Add Windows multithreading support for new toolchain
* Add splitting by maximally selected rank statistics for survival and regression forests
* Faster method for unordered factor splitting
* Add p-values for variable importance
* Runtime improvement for regression forests on classification data
* Bug fixes
##### Version 0.4.0
* Reduce memory usage of savest forest objects (changed child.nodeIDs interface)
* Add keep.inbag option to track in-bag counts
* Add option sample.fraction for fraction of sampled observations
* Add tree-wise split.select.weights
* Add predict.all option in predict() to get individual predictions for each tree for classification and regression
* Add case-specific random forests
* Add case weights (weighted bootstrapping or subsampling)
* Remove tuning functions, please use mlr or caret
* Catch error of outdated gcc not supporting C++11 completely
* Bug fixes
##### Version 0.3.0
* Allow the user to interrupt computation from R
* Transpose classification.table and rename to confusion.matrix
* Respect R seed for prediction
* Memory improvements for variable importance computation
* Fix bug: Probability prediction for single observations
* Fix bug: Results not identical when using alternative interface
##### Version 0.2.7
* Small fixes for Solaris compiler
##### Version 0.2.6
* Add C-index splitting
* Fix NA SNP handling
##### Version 0.2.5
* Fix matrix and gwaa alternative survival interface
* Version submitted to JSS
##### Version 0.2.4
* Small changes in documentation
##### Version 0.2.3
* Preallocate memory for splitting
##### Version 0.2.2
* Remove recursive splitting
##### Version 0.2.1
* Allow matrix as input data in R version
##### Version 0.2.0
* Fix prediction of classification forests in R
##### Version 0.1.9
* Speedup growing for continuous covariates
* Add memory save option to save memory for very large datasets (but slower)
* Remove memory mode option from R version since no performance gain
##### Version 0.1.8
* Fix problems when using Rcpp <0.11.4
##### Version 0.1.7
* Add option to split on unordered categorical covariates
##### Version 0.1.6
* Optimize memory management for very large survival forests
##### Version 0.1.5
* Set required Rcpp version to 0.11.2
* Fix large $call objects when using BatchJobs
* Add details and example on GenABEL usage to documentation
* Minor changes to documentation
##### Version 0.1.4
* Speedup for survival forests with continuous covariates
* R version: Generate seed from R. It is no longer necessary to set the
seed argument in ranger calls.
##### Version 0.1.3
* Windows support for R version (without multithreading)
##### Version 0.1.2
* Speedup growing of regression and probability prediction forests
* Prediction forests are now handled like regression forests: MSE used for
prediction error and permutation importance
* Fixed name conflict with randomForest package for "importance"
* Fixed a bug: prediction function is now working for probability
prediction forests
* Slot "predictions" for probability forests now contains class probabilities
* importance function is now working even if randomForest package is
loaded after ranger
* Fixed a bug: Split selection weights are now working as expected
* Small changes in documentation
|
