我正在尝试在 spark 中运行 Kmeans 集群，但我遇到了异常

Question

**Python Version = 2.6.6**

**numpy version = 1.3.0**  

**** The python file dokmeans.py is located in /home/cloudera****  






 Welcome to
          ____              __
         / __/__  ___ _____/ /__
        _\ \/ _ \/ _ `/ __/  '_/
       /__ / .__/\_,_/_/ /_/\_\   version 1.3.0
          /_/

    Using Python version 2.6.6 (r266:84292, Feb 22 2013 00:00:18)
    SparkContext available as sc, HiveContext available as sqlCtx.
    >>> exec(open('dokmeans.py').read())
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "<string>", line 5, in <module>
      File "/usr/lib/spark/python/pyspark/mllib/__init__.py", line 26, in <module>
        raise Exception("MLlib requires NumPy 1.4+")
    Exception: MLlib requires NumPy 1.4+
    >>> from pyspark.mllib.clustering import KMeans,KMeansModel
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "/usr/lib/spark/python/pyspark/mllib/__init__.py", line 26, in <module>
        raise Exception("MLlib requires NumPy 1.4+")
    Exception: MLlib requires NumPy 1.4+

Answer 1

mllib 代码中有一个错误，它不能正确解释 numpy 版本。它将 1.10 解释为 1.1，因此检查 numpy 版本失败。

请更改以下文件中的代码 /usr/lib/spark/python/pyspark/mllib/init.py

来自：

**if numpy.version.version < '1.4':**
raise Exception("MLlib requires NumPy 1.4+")

到：

ver = [int(x) for x in numpy.version.version.split('.')[:2]]
if ver < [1, 4]:
    raise Exception("MLlib requires NumPy 1.4+")

Answer 2

好吧，错误代码说明了一切。为了使用 MLlib，您需要安装 numpy 1.4。您已安装 1.3。