h2oai/h2o-3

Newer

Older

Raw Normal view History

100644 598 lines (458 sloc) 25.6 KB

Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	1	# Porting R Scripts
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	2
	3	This document outlines how to port R scripts written in previous versions of H2O (Nunes 2.8.6.2 or prior, also known as "H2O Classic") for compatibility with the new H2O 3.0 API. When upgrading from H2O to H2O 3.0, most functions are the same. However, there are some differences that will need to be resolved when porting any scripts that were originally created using H2O to H2O 3.0.
	4
	5	The original R script for H2O is listed first, followed by the updated script for H2O 3.0.
	6
	7	Some of the parameters have been renamed for consistency. For each algorithm, a table that describes the differences is provided.
	8
	9	For additional assistance within R, enter a question mark before the command (for example, `?h2o.glm`).
	10
	11	There is also a "shim" available that will review R scripts created with previous versions of H2O, identify deprecated or renamed parameters, and suggest replacements. For more information, refer to the repo [here](https://github.com/h2oai/h2o-dev/blob/d9693a97da939a2b77c24507c8b40a5992192489/h2o-r/h2o-package/R/shim.R).
	12
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	13	## Changes from H2O 2.8 to H2O 3.0
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	14
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	15	### `h2o.exec`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	16	The `h2o.exec` command is no longer supported. Any workflows using `h2o.exec` must be revised to remove this command. If the H2O 3.0 workflow contains any parameters or commands from H2O Classic, errors will result and the workflow will fail.
	17
	18	The purpose of `h2o.exec` was to wrap expressions so that they could be evaluated in a single `\Exec2` call. For example,
	19	`h2o.exec(fr[,1] + 2/fr[,3])`
	20	and
	21	`fr[,1] + 2/fr[,3]`
	22	produced the same results in H2O. However, the first example makes a single REST call and uses a single temp object, while the second makes several REST calls and uses several temp objects.
	23
	24	Due to the improved architecture in H2O 3.0, the need to use `h2o.exec` has been eliminated, as the expression can be processed by R as an "unwrapped" typical R expression.
	25
	26	Currently, the only known exception is when `factor` is used in conjunction with `h2o.exec`. For example, `h2o.exec(fr$myIntCol <- factor(fr$myIntCol))` would become `fr$myIntCol <- as.factor(fr$myIntCol)`
	27
	28	Note also that an array is not inside a string:
	29
	30	An int array is [1, 2, 3], not "[1, 2, 3]".
	31
	32	A String array is ["f00", "b4r"], not "[\"f00\", \"b4r\"]"
	33
	34	Only string values are enclosed in double quotation marks (`"`).
	35
	36	<a name="h2operf"></a>
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	37	### `h2o.performance`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	38
	39	To access any exclusively binomial output, use `h2o.performance`, optionally with the corresponding accessor. The accessor can only use the model metrics object created by `h2o.performance`. Each accessor is named for its corresponding field (for example, `h2o.AUC`, `h2o.gini`, `h2o.F1`). `h2o.performance` supports all current algorithms except for K-Means.
	40
	41	If you specify a data frame as a second parameter, H2O will use the specified data frame for scoring. If you do not specify a second parameter, the training metrics for the model metrics object are used.
	42
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	43	### `xval` and `validation` slots
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	44
	45	The `xval` slot has been removed, as `nfolds` is not currently supported.
	46
	47	The `validation` slot has been merged with the `model` slot.
	48
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	49	### Principal Components Regression (PCR)
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	50
	51	Principal Components Regression (PCR) has also been deprecated. To obtain PCR values, create a Principal Components Analysis (PCA) model, then create a GLM model from the scored data from the PCA model.
	52
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	53	### Saving and Loading Models
Jun 10, 2015 Update Recent Changes & Porting R Scripts doc with model save/restore
	54
	55	Saving and loading a model from R is supported in version 3.0.0.18 and later. H2O 3.0 uses the same binary serialization method as previous versions of H2O, but saves the model and its dependencies into a directory, with each object as a separate file. The `save_CV` option for available in previous versions of H2O has been deprecated, as `h2o.saveAll` and `h2o.loadAll` are not currently supported. The following commands are now supported:
	56
	57	- `h2o.saveModel`
	58	- `h2o.loadModel`
	59
	60
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	61
	62	Table of Contents
	63
	64	- [GBM](#GBM)
	65	- [GLM](#GLM)
	66	- [K-Means](#Kmeans)
	67	- [Deep Learning](#DL)
	68	- [Distributed Random Forest](#DRF)
	69
	70
	71
	72	<a name="GBM"></a>
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	73	## GBM
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	74
	75	N-fold cross-validation and grid search will be supported in a future version of H2O 3.0.
	76
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	77	### Renamed GBM Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	78
	79	The following parameters have been renamed, but retain the same functions:
	80
	81	H2O Classic Parameter Name \| H2O 3.0 Parameter Name
	82	-------------------\|-----------------------
	83	`data` \| `training_frame`
	84	`key` \| `model_id`
	85	`n.trees` \| `ntrees`
	86	`interaction.depth` \| `max_depth`
	87	`n.minobsinnode` \| `min_rows`
	88	`shrinkage` \| `learn_rate`
	89	`n.bins` \| `nbins`
	90	`validation` \| `validation_frame`
	91	`balance.classes` \| `balance_classes`
	92	`max.after.balance.size` \| `max_after_balance_size`
	93
	94
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	95	### Deprecated GBM Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	96
	97	The following parameters have been removed:
	98
	99	- `group_split`: Bit-set group splitting of categorical variables is now the default.
	100	- `importance`: Variable importances are now computed automatically and displayed in the model output.
	101	- `holdout.fraction`: The fraction of the training data to hold out for validation is no longer supported.
	102	- `grid.parallelism`: Specifying the number of parallel threads to run during a grid search is no longer supported. Grid search will be supported in a future version of H2O 3.0.
	103
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	104	### New GBM Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	105
	106	The following parameters have been added:
	107
	108	- `seed`: A random number to control sampling and initialization when `balance_classes` is enabled.
	109	- `score_each_iteration`: Display error rate information after each tree in the requested set is built.
Jun 17, 2015 Update Recent Changes; update docs for `build_tree_one_node` addition…
	110	- `build_tree_one_node`: Run on a single node to use fewer CPUs.
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	111
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	112	### GBM Algorithm Comparison
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	113
	114	H2O Classic \| H2O 3.0
	115	------------- \| -------------
	116	`h2o.gbm <- function(` \| `h2o.gbm <- function(`
	117	`x,` \|`x,`
	118	`y,` \|`y,`
	119	`data,` \| `training_frame,`
	120	`key = "",` \| `model_id,`
Dec 28, 2015 Updated with recent changes & fixed formatting
	121	\| `checkpoint`
	122	`distribution = 'multinomial',` \| `distribution = c("AUTO", "gaussian", "bernoulli", "multinomial", "poisson", "gamma", "tweedie"),`
	123	\| `tweedie_power = 1.5,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	124	`n.trees = 10,` \| `ntrees = 50`
	125	`interaction.depth = 5,` \| `max_depth = 5,`
	126	`n.minobsinnode = 10,` \| `min_rows = 10,`
Dec 28, 2015 Updated with recent changes & fixed formatting
	127	`shrinkage = 0.1,` \| `learn_rate = 0.1,`
	128	\| `sample_rate = 1`
	129	\| `col_sample_rate = 1`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	130	`n.bins = 20,`\| `nbins = 20,`
Dec 28, 2015 Updated with recent changes & fixed formatting
	131	\| `nbins_top_level,`
	132	\| `nbins_cats = 1024,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	133	`validation,` \| `validation_frame = NULL,`
	134	`balance.classes = FALSE` \| `balance_classes = FALSE,`
	135	`max.after.balance.size = 5,` \| `max_after_balance_size = 1,`
	136	\| `seed,`
Jun 17, 2015 Update Recent Changes; update docs for `build_tree_one_node` addition…
	137	\| `build_tree_one_node = FALSE,`
Dec 28, 2015 Updated with recent changes & fixed formatting
	138	\| `nfolds = 0,`
	139	\| `fold_column = NULL,`
	140	\| `fold_assignment = c("AUTO", "Random", "Modulo"),`
	141	\| `keep_cross_validation_predictions = FALSE,`
	142	\| `score_each_iteration = FALSE,`
	143	\| `stopping_rounds = 0,`
	144	\| `stopping_metric = c("AUTO", "deviance", "logloss", "MSE", "AUC", "r2", "misclassification"),`
	145	\| `stopping_tolerance = 0.001,`
	146	\| `offset_column = NULL,`
	147	\| `weights_column = NULL,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	148	`group_split = TRUE,` \|
	149	`importance = FALSE,` \|
	150	`holdout.fraction = 0,` \|
	151	`class.sampling.factors = NULL,` \|
	152	`grid.parallelism = 1)` \|
	153
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	154	### Output
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	155
	156	The following table provides the component name in H2O, the corresponding component name in H2O 3.0 (if supported), and the model type (binomial, multinomial, or all). Many components are now included in `h2o.performance`; for more information, refer to [(`h2o.performance`)](#h2operf).
	157
	158	H2O Classic \| H2O 3.0 \| Model Type
	159	------------- \| ------------- \| -------------
	160	`@model$priorDistribution`\|   \| `all`
	161	`@model$params` \| `@allparameters` \| `all`
	162	`@model$err` \| `@model$scoring_history` \| `all`
	163	`@model$classification` \|   \| `all`
	164	`@model$varimp` \| `@model$variable_importances` \| `all`
Jul 17, 2015 Updates for nightly build & other doc fixes
	165	`@model$confusion` \| `@model$training_metrics@metrics$cm$table` \| `binomial` and `multinomial`
	166	`@model$auc` \| `@model$training_metrics@metrics$AUC` \| `binomial`
	167	`@model$gini` \| `@model$training_metrics@metrics$Gini` \| `binomial`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	168	`@model$best_cutoff` \|   \| `binomial`
Jul 17, 2015 Updates for nightly build & other doc fixes
	169	`@model$F1` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$f1` \| `binomial`
	170	`@model$F2` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$f2` \| `binomial`
	171	`@model$accuracy` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$accuracy` \| `binomial`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	172	`@model$error` \|   \| `binomial`
Jul 17, 2015 Updates for nightly build & other doc fixes
	173	`@model$precision` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$precision` \| `binomial`
	174	`@model$recall` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$recall` \| `binomial`
	175	`@model$mcc` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$absolute_MCC` \| `binomial`
	176	`@model$max_per_class_err` \| currently replaced by `@model$training_metrics@metrics$thresholds_and_metric_scores$min_per_class_correct` \| `binomial`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	177
	178
	179
	180
	181
	182	---
	183
	184	<a name="GLM"></a>
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	185	## GLM
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	186
	187	N-fold cross-validation and grid search will be supported in a future version of H2O 3.0.
	188
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	189	### Renamed GLM Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	190
	191	The following parameters have been renamed, but retain the same functions:
	192
	193	H2O Classic Parameter Name \| H2O 3.0 Parameter Name
	194	-------------------\|-----------------------
	195	`data` \| `training_frame`
	196	`key` \| `model_id`
	197	`nlambda` \| `nlambdas`
	198	`lambda.min.ratio` \| `lambda_min_ratio`
	199	`iter.max` \| `max_iterations`
	200	`epsilon` \| `beta_epsilon`
	201
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	202	### Deprecated GLM Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	203
	204	The following parameters have been removed:
	205
	206	- `return_all_lambda`: A logical value indicating whether to return every model built during the lambda search. (may be re-added)
	207	- `higher_accuracy`: For improved accuracy, adjust the `beta_epsilon` value.
	208	- `strong_rules`: Discards predictors likely to have 0 coefficients prior to model building. (may be re-added as enabled by default)
	209	- `non_negative`: Specify a non-negative response. (may be re-added)
	210	- `variable_importances`: Variable importances are now computed automatically and displayed in the model output. They have been renamed to Normalized Coefficient Magnitudes.
	211	- `disable_line_search`: This parameter has been deprecated, as it was mainly used for testing purposes.
	212	- `offset`: Specify a column as an offset. (may be re-added)
	213	- `max_predictors`: Stops training the algorithm if the number of predictors exceeds the specified value. (may be re-added)
	214
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	215	### New GLM Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	216
	217	The following parameters have been added:
	218
	219	- `validation_frame`: Specify the validation dataset.
	220	- `solver`: Select IRLSM or LBFGS.
	221
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	222	### GLM Algorithm Comparison
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	223
	224
	225	H2O Classic \| H2O 3.0
	226	------------- \| -------------
	227	`h2o.glm <- function(` \| `h2o.startGLMJob <- function(`
	228	`x,` \| `x,`
	229	`y,` \| `y,`
	230	`data,` \|`training_frame,`
	231	`key = "",` \| `model_id,`
	232	\| `validation_frame`
	233	`iter.max = 100,` \| `max_iterations = 50,`
	234	`epsilon = 1e-4` \| `beta_epsilon = 0`
	235	`strong_rules = TRUE,` \|
	236	`return_all_lambda = FALSE,` \|
Jun 13, 2015 One more update for intercept (R docs)
Jun 13, 2015 One more update for intercept (R docs)	237	`intercept = TRUE,` \| `intercept = TRUE`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	238	`non_negative = FALSE,` \|
	239	\| `solver = c("IRLSM", "L_BFGS"),`
	240	`standardize = TRUE,` \| `standardize = TRUE,`
	241	`family,` \| `family = c("gaussian", "binomial", "poisson", "gamma", "tweedie"),`
	242	`link,` \| `link = c("family_default", "identity", "logit", "log", "inverse", "tweedie"),`
	243	`tweedie.p = ifelse(family == "tweedie",1.5, NA_real_)` \| `tweedie_variance_power = NaN,`
	244	\| `tweedie_link_power = NaN,`
	245	`alpha = 0.5,` \| `alpha = 0.5,`
	246	`prior = NULL` \| `prior = 0.0,`
	247	`lambda = 1e-5,` \| `lambda = 1e-05,`
	248	`lambda_search = FALSE,` \| `lambda_search = FALSE,`
	249	`nlambda = -1,` \| `nlambdas = -1,`
	250	`lambda.min.ratio = -1,` \| `lambda_min_ratio = 1.0,`
	251	`use_all_factor_levels = FALSE` \| `use_all_factor_levels = FALSE,`
	252	`nfolds = 0,` \| `nfolds = 0,`
	253	`beta_constraints = NULL,` \| `beta_constraint = NULL)`
	254	`higher_accuracy = FALSE,` \|
	255	`variable_importances = FALSE,` \|
	256	`disable_line_search = FALSE,` \|
	257	`offset = NULL,` \|
	258	`max_predictors = -1)` \|
	259
	260
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	261	### Output
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	262
	263
	264	The following table provides the component name in H2O, the corresponding component name in H2O 3.0 (if supported), and the model type (binomial, multinomial, or all). Many components are now included in `h2o.performance`; for more information, refer to [(`h2o.performance`)](#h2operf).
	265
	266	H2O Classic \| H2O 3.0 \| Model Type
	267	------------- \| ------------- \| -------------
	268	`@model$params` \| `@allparameters` \| `all`
	269	`@model$coefficients` \| `@model$coefficients` \| `all`
	270	`@model$nomalized_coefficients` \| `@model$coefficients_table$norm_coefficients` \| `all`
	271	`@model$rank` \| `@model$rank` \| `all`
	272	`@model$iter` \|`@model$iter` \| `all`
	273	`@model$lambda` \|   \| `all`
	274	`@model$deviance` \| `@model$residual_deviance` \| `all`
	275	`@model$null.deviance` \| `@model$null_deviance` \| `all`
	276	`@model$df.residual` \| `@model$residual_degrees_of_freedom` \| `all`
	277	`@model$df.null` \| `@model$null_degrees_of_freedom` \| `all`
	278	`@model$aic` \| `@model$AIC`\| `all`
	279	`@model$train.err` \|   \| `binomial`
	280	`@model$prior` \|   \| `binomial`
	281	`@model$thresholds` \| `@model$threshold` \| `binomial`
	282	`@model$best_threshold` \|   \| `binomial`
	283	`@model$auc` \| `@model$AUC` \| `binomial`
	284	`@model$confusion` \|   \| `binomial`
	285
	286	<a name="Kmeans"></a>
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	287	## K-Means
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	288
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	289	### Renamed K-Means Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	290
	291	The following parameters have been renamed, but retain the same functions:
	292
	293	H2O Classic Parameter Name \| H2O 3.0 Parameter Name
	294	-------------------\|-----------------------
	295	`data` \| `training_frame`
	296	`key` \| `model_id`
	297	`centers` \| `k`
	298	`cols` \| `x`
	299	`iter.max` \| `max_iterations`
	300	`normalize` \| `standardize`
	301
	302	Note In H2O, the `normalize` parameter was disabled by default. The `standardize` parameter is enabled by default in H2O 3.0 to provide more accurate results for datasets containing columns with large values.
	303
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	304	### New K-Means Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	305
	306	The following parameters have been added:
	307
	308	- `user` has been added as an additional option for the `init` parameter. Using this parameter forces the K-Means algorithm to start at the user-specified points.
	309	- `user_points`: Specify starting points for the K-Means algorithm.
	310
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	311	### K-Means Algorithm Comparison
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	312
	313	H2O Classic \| H2O 3.0
	314	------------- \| -------------
	315	`h2o.kmeans <- function(` \| `h2o.kmeans <- function(`
	316	`data,` \| `training_frame,`
	317	`cols = '',` \| `x,`
	318	`centers,` \| `k,`
	319	`key = "",` \| `model_id,`
	320	`iter.max = 10,` \| `max_iterations = 1000,`
	321	`normalize = FALSE,` \| `standardize = TRUE,`
	322	`init = "none",` \| `init = c("Furthest","Random", "PlusPlus"),`
	323	`seed = 0,` \| `seed)`
	324
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	325	### Output
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	326
	327
	328	The following table provides the component name in H2O and the corresponding component name in H2O 3.0 (if supported).
	329
	330	H2O Classic \| H2O 3.0
	331	------------- \| -------------
	332	`@model$params` \| `@allparameters`
	333	`@model$centers` \| `@model$centers`
	334	`@model$tot.withinss` \| `@model$tot_withinss`
	335	`@model$size` \| `@model$size`
	336	`@model$iter` \| `@model$iterations`
	337	\| `@model$_scoring_history`
	338	\| `@model$_model_summary`
	339
	340	---
	341
	342	<a name="DL"></a>
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	343	## Deep Learning
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	344
	345	N-fold cross-validation and grid search will be supported in a future version of H2O 3.0.
	346
	347	Note: If the results in the confusion matrix are incorrect, verify that `score_training_samples` is equal to 0. By default, only the first 10,000 rows are included.
	348
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	349	### Renamed Deep Learning Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	350
	351	The following parameters have been renamed, but retain the same functions:
	352
	353	H2O Classic Parameter Name \| H2O 3.0 Parameter Name
	354	-------------------\|-----------------------
	355	`data` \| `training_frame`
	356	`key` \| `model_id`
	357	`validation` \| `validation_frame`
	358	`class.sampling.factors` \| `class_sampling_factors`
	359	`override_with_best_model` \| `overwrite_with_best_model`
Jun 25, 2015 Update to Porting/Migration doc to fix MSE typo for DL; added info re…
	360	`dlmodel@model$valid_class_error` \| `@model$validation_metrics@$MSE`
	361
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	362
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	363	### Deprecated DL Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	364
	365	The following parameters have been removed:
	366
	367	- `classification`: Classification is now inferred from the data type.
	368	- `holdout_fraction`: Fraction of the training data to hold out for validation.
Jun 25, 2015 Update to Porting/Migration doc to fix MSE typo for DL; added info re…
	369	- `dlmodel@model$best_cutoff`: This output parameter has been removed.
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	370
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	371	### New DL Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	372
	373	The following parameters have been added:
	374
	375	- `export_weights_and_biases`: An additional option allowing users to export the raw weights and biases as H2O frames.
	376
	377	The following options for the `loss` parameter have been added:
	378
	379	- `absolute`: Provides strong penalties for mispredictions
	380	- `huber`: Can improve results for regression
	381
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	382	### DL Algorithm Comparison
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	383
	384	H2O Classic \| H2O 3.0
	385	------------- \| -------------
Dec 28, 2015 Updated with recent changes & fixed formatting
Dec 28, 2015 Updated with recent changes & fixed formatting	386	`h2o.deeplearning <- function(x,` \| `h2o.deeplearning (x, `
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	387	`y,` \| `y,`
	388	`data,` \| `training_frame,`
	389	`key = "",` \| `model_id = "",`
	390	`override_with_best_model,` \| `overwrite_with_best_model = true,`
	391	`classification = TRUE,` \|
Jul 17, 2015 Updates for nightly build & other doc fixes
Jul 17, 2015 Updates for nightly build & other doc fixes	392	`nfolds = 0,` \| `nfolds = 0`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	393	`validation,` \| `validation_frame,`
	394	`holdout_fraction = 0,` \|
	395	`checkpoint = " "` \| `checkpoint,`
	396	`autoencoder,` \| `autoencoder = false,`
	397	`use_all_factor_levels,` \| `use_all_factor_levels = true`
	398	`activation,` \| `_activation = c("Rectifier", "Tanh", "TanhWithDropout", "RectifierWithDropout", "Maxout", "MaxoutWithDropout"),`
	399	`hidden,` \| `hidden= c(200, 200),`
	400	`epochs,` \| `epochs = 10.0,`
	401	`train_samples_per_iteration,` \|`train_samples_per_iteration = -2,`
Dec 28, 2015 Updated with recent changes & fixed formatting
Dec 28, 2015 Updated with recent changes & fixed formatting	402	\| `target_ratio_comm_to_comp = 0.05`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	403	`seed,` \| `_seed,`
	404	`adaptive_rate,` \| `adaptive_rate = true,`
	405	`rho,` \| `rho = 0.99,`
Dec 28, 2015 Updated with recent changes & fixed formatting
Dec 28, 2015 Updated with recent changes & fixed formatting	406	`epsilon,` \| `epsilon = 1e-08,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	407	`rate,` \| `rate = .005,`
Dec 28, 2015 Updated with recent changes & fixed formatting
Dec 28, 2015 Updated with recent changes & fixed formatting	408	`rate_annealing,` \| `rate_annealing = 1e-06,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	409	`rate_decay,` \| `rate_decay = 1.0,`
	410	`momentum_start,` \| `momentum_start = 0,`
Dec 28, 2015 Updated with recent changes & fixed formatting
Dec 28, 2015 Updated with recent changes & fixed formatting	411	`momentum_ramp,` \| `momentum_ramp = 1e+06,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	412	`momentum_stable,` \| `momentum_stable = 0,`
	413	`nesterov_accelerated_gradient,` \| `nesterov_accelerated_gradient = true,`
	414	`input_dropout_ratio,` \| `input_dropout_ratio = 0.0,`
	415	`hidden_dropout_ratios,` \| `hidden_dropout_ratios,`
	416	`l1,` \| `l1 = 0.0,`
	417	`l2,` \| `l2 = 0.0,`
	418	`max_w2,` \| `max_w2 = Inf,`
	419	`initial_weight_distribution,` \| `initial_weight_distribution = c("UniformAdaptive","Uniform", "Normal"),`
	420	`initial_weight_scale,` \| `initial_weight_scale = 1.0,`
Dec 28, 2015 Updated with recent changes & fixed formatting
	421	`loss,` \| `loss = "Automatic", "CrossEntropy", "Quadratic", "Absolute", "Huber"),`
	422	\| `distribution = c("AUTO", "gaussian", "bernoulli", "multinomial", "poisson", "gamma", "tweedie", "laplace", "huber"),`
	423	\| `tweedie_power = 1.5,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	424	`score_interval,` \| `score_interval = 5,`
	425	`score_training_samples,` \| `score_training_samples = 10000l,`
	426	`score_validation_samples,` \| `score_validation_samples = 0l,`
	427	`score_duty_cycle,` \| `score_duty_cycle = 0.1,`
	428	`classification_stop,` \| `classification_stop = 0`
	429	`regression_stop,` \| `regression_stop = 1e-6,`
Dec 28, 2015 Updated with recent changes & fixed formatting
	430	\| `stopping_rounds = 5,`
	431	\| `stopping_metric = c("AUTO", "deviance", "logloss", "MSE", "AUC", "r2", "misclassification"),`
	432	\| `stopping_tolerance = 0,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	433	`quiet_mode,` \| `quiet_mode = false,`
	434	`max_confusion_matrix_size,` \| `max_confusion_matrix_size,`
	435	`max_hit_ratio_k,` \| `max_hit_ratio_k,`
	436	`balance_classes,` \| `balance_classes = false,`
	437	`class_sampling_factors,` \| `class_sampling_factors,`
	438	`max_after_balance_size,` \| `max_after_balance_size,`
	439	`score_validation_sampling,` \| `score_validation_sampling,`
	440	`diagnostics,` \| `diagnostics = true,`
	441	`variable_importances,` \| `variable_importances = false,`
	442	`fast_mode,` \| `fast_mode = true,`
	443	`ignore_const_cols,` \| `ignore_const_cols = true,`
	444	`force_load_balance,` \| `force_load_balance = true,`
	445	`replicate_training_data,` \| `replicate_training_data = true,`
	446	`single_node_mode,` \| `single_node_mode = false,`
	447	`shuffle_training_data,` \| `shuffle_training_data = false,`
	448	`sparse,` \| `sparse = false,`
	449	`col_major,` \| `col_major = false,`
Dec 28, 2015 Updated with recent changes & fixed formatting
Dec 28, 2015 Updated with recent changes & fixed formatting	450	`max_categorical_features,` \| `max_categorical_features,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	451	`reproducible)` \| `reproducible=FALSE,`
	452	`average_activation` \| `average_activation = 0,`
	453	\| `sparsity_beta = 0`
Dec 28, 2015 Updated with recent changes & fixed formatting
	454	\| `export_weights_and_biases=FALSE,`
	455	\| `offset_column = NULL,`
	456	\| `weights_column = NULL,`
	457	\| `nfolds = 0,`
	458	\| `fold_column = NULL,`
	459	\| `fold_assignment = c("AUTO", "Random", "Modulo"),`
	460	\| `keep_cross_validation_predictions = FALSE)`
	461
	462
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	463	### Output
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	464
	465
	466	The following table provides the component name in H2O, the corresponding component name in H2O 3.0 (if supported), and the model type (binomial, multinomial, or all). Many components are now included in `h2o.performance`; for more information, refer to [(`h2o.performance`)](#h2operf).
	467
	468	H2O Classic \| H2O 3.0 \| Model Type
	469	------------- \| ------------- \| -------------
	470	`@model$priorDistribution`\|   \| `all`
	471	`@model$params` \| `@allparameters` \| `all`
Jul 17, 2015 Updates for nightly build & other doc fixes
Jul 17, 2015 Updates for nightly build & other doc fixes	472	`@model$train_class_error` \| `@model$training_metrics@metrics@$MSE` \| `all`
Jun 25, 2015 Update to Porting/Migration doc to fix MSE typo for DL; added info re…
	473	`@model$valid_class_error` \| `@model$validation_metrics@$MSE` \| `all`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	474	`@model$varimp` \| `@model$_variable_importances` \| `all`
Jul 17, 2015 Updates for nightly build & other doc fixes
Jul 17, 2015 Updates for nightly build & other doc fixes	475	`@model$confusion` \| `@model$training_metrics@metrics$cm$table` \| `binomial` and `multinomial`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	476	`@model$train_auc` \| `@model$train_AUC` \| `binomial`
	477	\| `@model$_validation_metrics` \| `all`
	478	\| `@model$_model_summary` \| `all`
	479	\| `@model$_scoring_history` \| `all`
	480
	481
	482	---
	483
	484	<a name="DRF"></a>
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	485	## Distributed Random Forest
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	486
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	487	### Changes to DRF in H2O 3.0
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	488
Jun 11, 2015 Misc. typo fixes
Jun 11, 2015 Misc. typo fixes	489	Distributed Random Forest (DRF) was represented as `h2o.randomForest(type="BigData", ...)` in H2O Classic. In H2O Classic, SpeeDRF (`type="fast"`) was not as accurate, especially for complex data with categoricals, and did not address regression problems. DRF (`type="BigData"`) was at least as accurate as SpeeDRF (`type="fast"`) and was the only algorithm that scaled to big data (data too large to fit on a single node).
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	490	In H2O 3.0, our plan is to improve the performance of DRF so that the data fits on a single node (optimally, for all cases), which will make SpeeDRF obsolete. Ultimately, the goal is provide a single algorithm that provides the "best of both worlds" for all datasets and use cases.
Jun 15, 2015 Update team list, add note re: multi-file zip, h2o.predict for DRF re…
	491	Please note that H2O does not currently support the ability to specify the number of trees when using `h2o.predict` for a DRF model.
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	492
	493	Note: H2O 3.0 only supports DRF. SpeeDRF is no longer supported. The functionality of DRF in H2O 3.0 is similar to DRF functionality in H2O.
	494
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	495	### Renamed DRF Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	496
	497	The following parameters have been renamed, but retain the same functions:
	498
	499	H2O Classic Parameter Name \| H2O 3.0 Parameter Name
	500	-------------------\|-----------------------
	501	`data` \| `training_frame`
	502	`key` \| `model_id`
	503	`validation` \| `validation_frame`
	504	`sample.rate` \| `sample_rate`
	505	`ntree` \| `ntrees`
	506	`depth` \| `max_depth`
	507	`balance.classes` \| `balance_classes`
	508	`score.each.iteration` \| `score_each_iteration`
	509	`class.sampling.factors` \| `class_sampling_factors`
	510	`nodesize` \| `min_rows`
	511
	512
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	513	### Deprecated DRF Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	514
	515	The following parameters have been removed:
	516
	517	- `classification`: This is now automatically inferred from the response type. To achieve classification with a 0/1 response column, explicitly convert the response to a factor (`as.factor()`).
	518	- `importance`: Variable importances are now computed automatically and displayed in the model output.
	519	- `holdout.fraction`: Specifying the fraction of the training data to hold out for validation is no longer supported.
	520	- `doGrpSplit`: The bit-set group splitting of categorical variables is now the default.
	521	- `verbose`: Infonrmation about tree splits and extra statistics is now included automatically in the stdout.
	522	- `oobee`: The out-of-bag error estimate is now computed automatically (if no validation set is specified).
	523	- `stat.type`: This parameter was used for SpeeDRF, which is no longer supported.
	524	- `type`: This parameter was used for SpeeDRF, which is no longer supported.
	525
Jun 25, 2015 Update to Porting/Migration doc to fix MSE typo for DL; added info re…
	526
	527
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	528	### New DRF Parameters
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	529
	530	The following parameter has been added:
	531
	532	- `build_tree_one_node`: Run on a single node to use fewer CPUs.
	533
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	534	### DRF Algorithm Comparison
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	535
	536	H2O Classic \| H2O 3.0
	537	------------- \| -------------
	538	`h2o.randomForest <- function(x,` \| `h2o.randomForest <- function(`
	539	`x,` \| `x,`
	540	`y,` \| `y,`
	541	`data,` \| `training_frame,`
	542	`key="",` \| `model_id,`
	543	`validation,` \| `validation_frame,`
	544	`mtries = -1,` \| `mtries = -1,`
May 21, 2015 Update default value for `sample_rate` (per Arno)
	545	`sample.rate=2/3,` \| `sample_rate = 0.632,`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	546	\| `build_tree_one_node = FALSE,`
	547	`ntree=50` \| `ntrees=50,`
	548	`depth=20,` \| `max_depth = 20,`
	549	\| `min_rows = 1,`
	550	`nbins=20,` \| `nbins = 20,`
	551	`balance.classes = FALSE,` \| `balance_classes = FALSE,`
	552	`score.each.iteration = FALSE,` \| `score_each_iteration = FALSE,`
	553	`seed = -1,` \| `seed`
	554	`nodesize = 1,` \|
	555	`classification=TRUE,` \|
	556	`importance=FALSE,` \|
	557	`nfolds=0,` \|
	558	`holdout.fraction = 0,` \|
	559	`max.after.balance.size = 5,` \| `max_after_balance_size)`
	560	`class.sampling.factors = NULL,` \|
	561	`doGrpSplit = TRUE,` \|
	562	`verbose = FALSE,` \|
	563	`oobee = TRUE,` \|
	564	`stat.type = "ENTROPY",` \|
	565	`type = "fast")` \|
	566
	567
Jun 23, 2017 PUBDEV-4213: Updates to markdown syntax (#1307 )
	568	### Output
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	569
	570
	571	The following table provides the component name in H2O, the corresponding component name in H2O 3.0 (if supported), and the model type (binomial, multinomial, or all). Many components are now included in `h2o.performance`; for more information, refer to [(`h2o.performance`)](#h2operf).
	572
	573	H2O Classic \| H2O 3.0 \| Model Type
	574	------------- \| ------------- \| -------------
	575	`@model$priorDistribution`\|   \| `all`
	576	`@model$params` \| `@allparameters` \| `all`
	577	`@model$mse` \| `@model$scoring_history` \| `all`
	578	`@model$forest` \| `@model$model_summary` \| `all`
	579	`@model$classification` \|   \| `all`
	580	`@model$varimp` \| `@model$variable_importances` \| `all`
Jul 17, 2015 Updates for nightly build & other doc fixes
	581	`@model$confusion` \| `@model$training_metrics@metrics$cm$table` \| `binomial` and `multinomial`
	582	`@model$auc` \| `@model$training_metrics@metrics$AUC` \| `binomial`
	583	`@model$gini` \| `@model$training_metrics@metrics$Gini` \| `binomial`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	584	`@model$best_cutoff` \|   \| `binomial`
Jul 17, 2015 Updates for nightly build & other doc fixes
	585	`@model$F1` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$f1` \| `binomial`
	586	`@model$F2` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$f2` \| `binomial`
	587	`@model$accuracy` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$accuracy` \| `binomial`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	588	`@model$Error` \| `@model$Error` \| `binomial`
Jul 17, 2015 Updates for nightly build & other doc fixes
	589	`@model$precision` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$precision` \| `binomial`
	590	`@model$recall` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$recall` \| `binomial`
	591	`@model$mcc` \| `@model$training_metrics@metrics$thresholds_and_metric_scores$absolute_MCC` \| `binomial`
	592	`@model$max_per_class_err` \| currently replaced by `@model$training_metrics@metrics$thresholds_and_metric_scores$min_per_class_correct` \| `binomial`
May 16, 2015 Restore Porting R Scripts doc, add Migration Guide
	593
	594
	595
	596
	597