Spark MLlib Statistics统计
1、Spark MLlib Statistics统计
Spark Mllib 统计模块代码结构如下：
1.1 列统计汇总
计算每列最大值、最小值、平均值、方差值、L1范数、L2范数。
//读取数据，转换成RDD[Vector]类型
val data_path = "/home/jb-huangmeiling/sample_stat.txt"
val data = sc.textFile(data_path).map(_.split("\t")).map(f => f.map(f => f.toDouble))
val data1 = data.map(f => Vectors.dense(f))
//计算每列最大值、最小值、平均值、方差值、L1范数、L2范数
val stat1 = Statistics.colStats(data1)
stat1.max
stat1.min
stat1.mean
stat1.variance
stat1.normL1
stat1.normL2
执行结果：
数据
| 1 |
2 |
3 |
4 |
5 |
| 6 |
7 |
1 |
5 |
9 |
| 3 |
5 |
6 |
3 |
1 |
| 3 |
1 |
1 |
5 |
6 |
scala> data1.collect
res19: Array[org.apache.spark.mllib.linalg.Vector] = Array([1.0,2.0,3.0,4.0,5.0], [6.0,7.0,1.0,5.0,9.0], [3.0,5.0,6.0,3.0,1.0], [3.0,1.0,1.0,5.0,6.0])
scala> stat1.max
res20: org.apache.spark.mllib.linalg.Vector = [6.0,7.0,6.0,5.0,9.0]
scala> stat1.min
res21: org.apache.spark.mllib.linalg.Vector = [1.0,1.0,1.0,3.0,1.0]
scala> stat1.mean
res22: org.apache.spark.mllib.linalg.Vector = [3.25,3.75,2.75,4.25,5.25]
scala> stat1.variance
res23: org.apache.spark.mllib.linalg.Vector = [4.25,7.583333333333333,5.583333333333333,0.9166666666666666,10.916666666666666]
scala> stat1.normL1
res24: org.apache.spark.mllib.linalg.Vector = [13.0,15.0,11.0,17.0,21.0]
scala> stat1.normL2
res25: org.apache.spark.mllib.linalg.Vector = [7.416198487095663,8.888194417315589,6.855654600401044,8.660254037844387,11.958260743101398]
1.2 相关系数
Pearson相关系数表达的是两个数值变量的线性相关性, 它一般适用于正态分布。其取值范围是[-1, 1], 当取值为0表示不相关，取值为(0~-1]表示负相关，取值为(0, 1]表示正相关。
Spearman相关系数也用来表达两个变量的相关性，但是它没有Pearson相关系数对变量的分布要求那么严格，另外Spearman相关系数可以更好地用于测度变量的排序关系。其计算公式为：
//计算pearson系数、spearman相关系数
val corr1 = Statistics.corr(data1, "pearson")
val corr2 = Statistics.corr(data1, "spearman")
val x1 = sc.parallelize(Array(1.0, 2.0, 3.0, 4.0))
val y1 = sc.parallelize(Array(5.0, 6.0, 6.0, 6.0))
val corr3 = Statistics.corr(x1, y1, "pearson")
scala> corr1
res6: org.apache.spark.mllib.linalg.Matrix =
1.0 0.7779829610026362 -0.39346431156047523 ... (5 total)
0.7779829610026362 1.0 0.14087521363240252 ...
-0.39346431156047523 0.14087521363240252 1.0 ...
0.4644203640128242 -0.09482093118615205 -0.9945577827230707 ...
0.5750122832421579 0.19233705001984078 -0.9286374704669208 ...
scala> corr2
res7: org.apache.spark.mllib.linalg.Matrix =
1.0 0.632455532033675 -0.5000000000000001 ... (5 total)
0.632455532033675 1.0 0.10540925533894883 ...
-0.5000000000000001 0.10540925533894883 1.0 ...
0.5000000000000001 -0.10540925533894883 -1.0000000000000002 ...
0.6324555320336723 0.20000000000000429 -0.9486832980505085 ...
scala> corr3
res8: Double = 0.7745966692414775
1.3 假设检验
MLlib当前支持用于判断拟合度或者独立性的Pearson卡方(chi-squared ( χ2) )检验。不同的输入类型决定了是做拟合度检验还是独立性检验。拟合度检验要求输入为Vector, 独立性检验要求输入是Matrix。
//卡方检验
val v1 = Vectors.dense(43.0, 9.0)
val v2 = Vectors.dense(44.0, 4.0)
val c1 = Statistics.chiSqTest(v1, v2)
执行结果：
c1: org.apache.spark.mllib.stat.test.ChiSqTestResult =
Chi squared test summary:
method: pearson
degrees of freedom = 1
statistic = 5.482517482517483
pValue = 0.01920757707591003
Strong presumption against null hypothesis: observed follows the same distribution as expected..
结果返回：统计量：pearson、自由度：1、值：5.48、概率：0.019。