-
Notifications
You must be signed in to change notification settings - Fork 3
/
Stacking_H2O.R
166 lines (146 loc) · 7.24 KB
/
Stacking_H2O.R
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
#=============================================================
# Load required packages
#=============================================================
require(h2oEnsemble)
require(ggplot2)
require(h2o)
#=============================================================
# Init H2O (connect to a running H2O cluster)
#=============================================================
h2o.init(port = 54321)
#h2o.removeAll()
#=============================================================
# Load data
#=============================================================
data_frame <-
h2o.importFile(
path = "http://www.dataminingconsultant.com/data/churn.txt",
sep = ",",
destination_frame = "data_frame")
# remove special characters from column names
colnames(data_frame) <- gsub(" ", "_", trimws(gsub("[[:punct:]]", " ", names(data_frame))))
#=============================================================
# Force classification
#=============================================================
data_frame$Churn <- as.factor(data_frame$Churn)
#=============================================================
# Split data into training and validation
#=============================================================
split_df <- h2o.splitFrame(data_frame, 0.7,
destination_frames = c("train_frame","valid_frame"),
seed=2016)
train_frame <- split_df[[1]]
valid_frame <- split_df[[2]]
#=============================================================
# Target and predictors
#=============================================================
y <- "Churn"
x <- setdiff(names(data_frame), y)
#=============================================================
# Specify base learners
#=============================================================
h2o.gbm.1 <-
function(...,
ntrees = 50, learn_rate = 0.03,
max_depth = 3, col_sample_rate = 0.65,
sample_rate = 0.8, seed = seed,
stopping_rounds = 150, stopping_metric = "AUC",
stopping_tolerance = 0.0005)
h2o.gbm.wrapper(...,
ntrees = ntrees, learn_rate = learn_rate,
max_depth = max_depth, col_sample_rate = col_sample_rate,
sample_rate = sample_rate, seed = seed,
stopping_rounds = stopping_rounds, stopping_metric = stopping_metric,
stopping_tolerance = stopping_tolerance)
h2o.gbm.2 <-
function(...,
ntrees = 25, learn_rate = 0.03,
max_depth = 5, col_sample_rate = 0.65,
sample_rate = 0.8, seed = seed,
stopping_rounds = 150, stopping_metric = "AUC",
stopping_tolerance = 0.0005)
h2o.gbm.wrapper(...,
ntrees = ntrees, learn_rate = learn_rate,
max_depth = max_depth, col_sample_rate = col_sample_rate,
sample_rate = sample_rate, seed = seed,
stopping_rounds = stopping_rounds, stopping_metric = stopping_metric,
stopping_tolerance = stopping_tolerance)
h2o.drf.1 <-
function(...,
binomial_double_trees = TRUE,
ntrees = 50, max_depth = 3,
sample_rate = 0.8, stopping_rounds = 150,
stopping_metric = "AUC", stopping_tolerance = 0.0005)
h2o.randomForest.wrapper(...,
binomial_double_trees = binomial_double_trees,
ntrees = ntrees, max_depth = max_depth,
sample_rate = sample_rate, stopping_rounds = stopping_rounds,
stopping_metric = stopping_metric, stopping_tolerance = stopping_tolerance)
h2o.drf.2 <-
function(...,
binomial_double_trees = TRUE,
ntrees = 20, max_depth = 5,
sample_rate = 0.8, stopping_rounds = 150,
stopping_metric = "AUC", stopping_tolerance = 0.0005)
h2o.randomForest.wrapper(...,
binomial_double_trees = binomial_double_trees,
ntrees = ntrees, max_depth = max_depth,
sample_rate = sample_rate, stopping_rounds = stopping_rounds,
stopping_metric = stopping_metric, stopping_tolerance = stopping_tolerance)
h2o.dl.1 <-
function(...,
activation = "TanhWithDropout", epochs = 600,
hidden = c(10,10), hidden_dropout_ratios = c(0.2,0.2),
input_dropout_ratio = 0.2, score_interval = 0.0001)
h2o.deeplearning.wrapper(...,
activation = activation, epochs = epochs ,
hidden = hidden, hidden_dropout_ratios = hidden_dropout_ratios,
input_dropout_ratio = input_dropout_ratio, score_interval = score_interval)
h2o.dl.2 <-
function(...,
activation = "TanhWithDropout", epochs = 600,
hidden = c(50,50), hidden_dropout_ratios = c(0.5,0.5),
input_dropout_ratio = 0.2, score_interval = 0.0001)
h2o.deeplearning.wrapper(...,
activation = activation, epochs = epochs ,
hidden = hidden, hidden_dropout_ratios = hidden_dropout_ratios,
input_dropout_ratio = input_dropout_ratio, score_interval = score_interval)
#=============================================================
# Create base learner vector
#=============================================================
learner <- c("h2o.gbm.1", "h2o.gbm.2",
"h2o.drf.1", "h2o.drf.2",
"h2o.dl.1", "h2o.dl.2")
#=============================================================
# Specify meta-learner
#=============================================================
# metalearner <- "SL.glm"
metalearner <- "h2o.glm.wrapper"
#=============================================================
# Fit Stacked learners
#=============================================================
family <- "binomial"
ensemble.fit <- h2o.ensemble(
x = x,
y = y,
training_frame = train_frame,
family = family,
learner = learner,
metalearner = metalearner,
cvControl = list(V = 5, shuffle = TRUE))
#=============================================================
# Save the models if you want
#=============================================================
# h2o.save_ensemble(fit, path = "ens_models/", export_levelone = TRUE)
perf <- h2o.ensemble_performance(
object = ensemble.fit,
newdata = valid_frame,
score_base_models = TRUE)
#=============================================================
# Print stacked fit AUC score
#=============================================================
print(perf, metric = "AUC")
#=============================================================
# Print stacked fit MSE score
#=============================================================
print(perf, metric = "MSE")