Aggregation.java
3.62 KB
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
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.SparkSession;
import org.apache.spark.sql.expressions.Window;
import org.apache.spark.sql.expressions.WindowSpec;
import static org.apache.spark.sql.functions.*;
import static org.apache.spark.sql.functions.lit;
import static org.apache.spark.sql.functions.when;
public class Aggregation {
public static void main(String[] args) throws Exception {
//Create Session
SparkSession spark = SparkSession
.builder()
.appName("Detecting Fraud Clicks")
.master("local")
.getOrCreate();
// Aggregation
Aggregation agg = new Aggregation();
Dataset<Row> dataset = agg.loadCSVDataSet("./train_sample.csv", spark);
dataset = agg.changeTimestempToLong(dataset);
dataset = agg.averageValidClickCount(dataset);
dataset = agg.clickTimeDelta(dataset);
dataset = agg.countClickInTenMinutes(dataset);
//test
dataset.where("ip == '5348' and app == '19'").show(10);
}
private Dataset<Row> loadCSVDataSet(String path, SparkSession spark){
// Read SCV to DataSet
return spark.read().format("csv")
.option("inferSchema", "true")
.option("header", "true")
.load(path);
}
private Dataset<Row> changeTimestempToLong(Dataset<Row> dataset){
// cast timestamp to long
Dataset<Row> newDF = dataset.withColumn("utc_click_time", dataset.col("click_time").cast("long"));
newDF = newDF.withColumn("utc_attributed_time", dataset.col("attributed_time").cast("long"));
newDF = newDF.drop("click_time").drop("attributed_time");
return newDF;
}
private Dataset<Row> averageValidClickCount(Dataset<Row> dataset){
// set Window partition by 'ip' and 'app' order by 'utc_click_time' select rows between 1st row to current row
WindowSpec w = Window.partitionBy("ip", "app")
.orderBy("utc_click_time")
.rowsBetween(Window.unboundedPreceding(), Window.currentRow());
// aggregation
Dataset<Row> newDF = dataset.withColumn("cum_count_click", count("utc_click_time").over(w));
newDF = newDF.withColumn("cum_sum_attributed", sum("is_attributed").over(w));
newDF = newDF.withColumn("avg_valid_click_count", col("cum_sum_attributed").divide(col("cum_count_click")));
newDF = newDF.drop("cum_count_click", "cum_sum_attributed");
return newDF;
}
private Dataset<Row> clickTimeDelta(Dataset<Row> dataset){
WindowSpec w = Window.partitionBy ("ip")
.orderBy("utc_click_time");
Dataset<Row> newDF = dataset.withColumn("lag(utc_click_time)", lag("utc_click_time",1).over(w));
newDF = newDF.withColumn("click_time_delta", when(col("lag(utc_click_time)").isNull(),
lit(0)).otherwise(col("utc_click_time")).minus(when(col("lag(utc_click_time)").isNull(),
lit(0)).otherwise(col("lag(utc_click_time)"))));
newDF = newDF.drop("lag(utc_click_time)");
return newDF;
}
private Dataset<Row> countClickInTenMinutes(Dataset<Row> dataset){
WindowSpec w = Window.partitionBy("ip")
.orderBy("utc_click_time")
.rangeBetween(Window.currentRow(),Window.currentRow()+600);
Dataset<Row> newDF = dataset.withColumn("count_click_in_ten_mins",
(count("utc_click_time").over(w)).minus(1)); //TODO 본인것 포함할 것인지 정해야함.
return newDF;
}
}