-
Notifications
You must be signed in to change notification settings - Fork 0
/
res_train_21746332_180423_0.9730.R
172 lines (122 loc) · 5.66 KB
/
res_train_21746332_180423_0.9730.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
166
167
168
169
170
171
172
if (!require("pacman")) install.packages("pacman")
pacman::p_load(knitr, tidyverse, highcharter, data.table, lubridate, pROC, tictoc, DescTools, lightgbm)
set.seed(84)
options(scipen = 9999, warn = -1, digits= 4)
train <- fread("train_reduce_2.csv",
col.names =c("ip", "app", "device", "os", "channel", "click_time",
"is_attributed", "ip_nextClick", "ip_app_nextClick", "ip_channel_nextClick", "ip_os_nextClick"),
showProgress = FALSE)
invisible(gc())
most_freq_hours_in_test_data <- c("4","5","9","10","13","14")
least_freq_hours_in_test_data <- c("6","11","15")
train$ip_nextClick <- NULL
train$ip_os_nextClick <- NULL
train[, UsrappCount:=.N, by=list(ip,app,device,os)]
train[, app_f := .N, by = "app"]
train[, ip_dev_f := .N, by = "ip,device"]
train[, ip_os_f := .N, by = "ip,os"]
train[, ip_app_f := .N, by = "ip,app"]
train[, channel_f := .N, by = "channel"]
train <- train %>% mutate(wday = Weekday(click_time),
hour = hour(click_time),
in_test_hh = ifelse(hour %in% most_freq_hours_in_test_data, 1,
ifelse(hour %in% least_freq_hours_in_test_data, 3, 2)))
train <- train %>%
select(-c(click_time)) %>%
add_count(ip, wday, in_test_hh) %>% rename("nip_day_test_hh" = n) %>%
select(-c(in_test_hh)) %>%
add_count(ip) %>% rename("n_rowip" = n) %>%
add_count(os, device) %>% rename("n_os_dev" = n) %>%
add_count(os, app, channel) %>% rename("n_os_app_chan" = n) %>%
add_count(app, channel) %>% rename("n_app_chan" = n) %>%
add_count(ip, wday, hour) %>% rename("n_ip" = n) %>%
add_count(ip, wday, hour, os) %>% rename("n_ip_os" = n) %>%
add_count(ip, wday, hour, app) %>% rename("n_ip_app" = n) %>%
add_count(ip, wday, hour, app, os) %>% rename("n_ip_app_os" = n) %>%
add_count(app, wday, hour) %>% rename("n_app" = n) %>%
add_count(ip, device, wday, hour) %>% rename("nip_dev_d_h" = n) %>%
select(-c(wday)) %>% select(-c(ip))
gc()
train[is.na(train)] <- 0
gc()
library(caret)
set.seed(71)
train_part <- createDataPartition(train$is_attributed, p = 0.7, list = FALSE)
categorical_features = c("app", "os", "channel", "hour")
dtrain = lgb.Dataset(data.matrix(train[train_part,] %>% select(-is_attributed)),
label = as.numeric(train[train_part,]$is_attributed),
categorical_feature = categorical_features)
dvalid = lgb.Dataset(data.matrix(train[-train_part,] %>% select(-is_attributed)),
label = as.numeric(train[-train_part,]$is_attributed),
categorical_feature = categorical_features)
rm(train)
invisible(gc())
params = list(objective = "binary",
metric = "auc",
learning_rate= 0.1,
num_leaves= 7,
max_depth= 3,
min_child_samples= 100,
max_bin= 255, # RAM dependent as per LightGBM documentation
subsample= 0.7,
subsample_freq= 1,
colsample_bytree= 0.7,
min_child_weight= 0,
min_split_gain= 0,
scale_pos_weight=99.75) # calculated for this dataset
set.seed(71)
model <- lgb.train(params, dtrain, valids = list(validation = dvalid), nthread = 7,
nrounds = 3000, verbose= 1, early_stopping_rounds = 50, eval_freq = 25)
kable(lgb.importance(model, percentage = TRUE))
#Predict
test <- fread("test.csv", colClasses = list(numeric=2:6), showProgress = FALSE)
sub <- data.table(click_id = test$click_id, is_attributed = NA)
test$click_id <- NULL
invisible(gc())
test$ip_nextClick <- NULL
test$ip_os_nextClick <- NULL
test[, UsrappRank:=1:.N, by=list(ip,app,device,os)]
test[, UsrappCount:=.N, by=list(ip,app,device,os)]
test[, UsrappNewness:=1:.N, by=list(ip,app,device,os)]
test[, UsrCount:=.N, by=list(ip,device,os)]
test[, UsrNewness:=1:.N, by=list(ip,device,os)]
test[, app_f := .N, by = "app"]
test[, ip_dev_f := .N, by = "ip,device"]
test[, ip_os_f := .N, by = "ip,os"]
test[, ip_app_f := .N, by = "ip,app"]
test[, channel_f := .N, by = "channel"]
test <- test %>% mutate(wday = Weekday(click_time),
hour = hour(click_time),
in_test_hh = ifelse(hour %in% most_freq_hours_in_test_data, 1,
ifelse(hour %in% least_freq_hours_in_test_data, 3, 2)))
test <- test %>%
select(-c(click_time)) %>%
add_count(ip, wday, in_test_hh) %>% rename("nip_day_test_hh" = n) %>%
select(-c(in_test_hh)) %>%
add_count(ip) %>% rename("n_rowip" = n) %>%
add_count(os, device) %>% rename("n_os_dev" = n) %>%
add_count(os, app, channel) %>% rename("n_os_app_chan" = n) %>%
add_count(ip, device) %>% rename("n_ip_dev" = n) %>%
add_count(app, channel) %>% rename("n_app_chan" = n) %>%
add_count(ip, wday, hour) %>% rename("n_ip" = n) %>%
add_count(ip, wday, hour, os) %>% rename("n_ip_os" = n) %>%
add_count(ip, wday, hour, app) %>% rename("n_ip_app" = n) %>%
add_count(ip, wday, hour, app, os) %>% rename("n_ip_app_os" = n) %>%
add_count(app, wday, hour) %>% rename("n_app" = n) %>%
add_count(device, hour) %>% rename("ndev_h" = n) %>%
add_count(ip, device, wday, hour) %>% rename("nip_dev_d_h" = n) %>%
add_count(channel, hour, device) %>% rename("nchan_dev_h" = n) %>%
select(-c(wday)) %>% select(-c(ip))
gc()
test[is.na(test)] <- 0
gc()
preds <- predict(model, data = data.matrix(test[, colnames(test)]), n = model$best_iter)
preds <- as.data.frame(preds)
sub$is_attributed <- NULL
sub$is_attributed <- preds
rm(test)
invisible(gc())
fwrite(sub, "res_train_21746332_180423.csv")
summary(preds$preds)
kable(lgb.importance(model, percentage = TRUE))
#0.9730