Spark 机器学习实践 ：Iris数据集的分类

今天试用了一下Spark的机器学习，体验如下：

第一步，导入数据

我们使用Iris数据集，做一个分类，首先要把csv文件导入。这里用到了spark的csv包，不明白为什么这么常见的功能不是内置的，还需要额外加载。

--packages com.databricks:spark-csv_2.11:1.4.0

from pyspark.sql import SQLContext
sqlContext = SQLContext(sc)
df = sqlContext.read.format('com.databricks.spark.csv')
    .options(header='true', inferschema='true')
    .load('iris.csv')
# Displays the content of the DataFrame to stdout
df.show()

结果如下：

+-----+------------+-----------+------------+-----------+-------+
|rowid|Sepal.Length|Sepal.Width|Petal.Length|Petal.Width|Species|
+-----+------------+-----------+------------+-----------+-------+
|    1|         5.1|        3.5|         1.4|        0.2| setosa|
|    2|         4.9|        3.0|         1.4|        0.2| setosa|
|    3|         4.7|        3.2|         1.3|        0.2| setosa|
|    4|         4.6|        3.1|         1.5|        0.2| setosa|
|    5|         5.0|        3.6|         1.4|        0.2| setosa|
|    6|         5.4|        3.9|         1.7|        0.4| setosa|
|    7|         4.6|        3.4|         1.4|        0.3| setosa|
|    8|         5.0|        3.4|         1.5|        0.2| setosa|
|    9|         4.4|        2.9|         1.4|        0.2| setosa|
|   10|         4.9|        3.1|         1.5|        0.1| setosa|
|   11|         5.4|        3.7|         1.5|        0.2| setosa|
|   12|         4.8|        3.4|         1.6|        0.2| setosa|
|   13|         4.8|        3.0|         1.4|        0.1| setosa|
|   14|         4.3|        3.0|