PySpark：AttributeError：“PipelineModel”对象没有属性“clusterCenters”

【问题标题】：PySpark: AttributeError: 'PipelineModel' object has no attribute 'clusterCenters'PySpark：AttributeError：“PipelineModel”对象没有属性“clusterCenters”
【发布时间】：2020-06-10 07:51:21
【问题描述】：

我用 Pypsark 创建了一个 kmeans 算法。现在，我还想提取聚类中心。如何将其包含在管道中？这是我到目前为止的代码，但它给我一个错误'AttributeError：'PipelineModel'对象没有属性'clusterCenters'。怎么解决？

#### model K-Means ###

from pyspark.ml.clustering import KMeans, KMeansModel

kmeans = KMeans() \
          .setK(3) \
          .setFeaturesCol("scaledFeatures")\
          .setPredictionCol("cluster")

# Chain indexer and tree in a Pipeline
pipeline = Pipeline(stages=[kmeans])

model = pipeline.fit(matrix_normalized)

cluster = model.transform(matrix_normalized)

#get cluster centers
centers = model.clusterCenters()

【问题讨论】：

可能是model[0].clusterCenter()？
不幸的是，这给了我 TypeError: 'PipelineModel' object is not subscriptable
对不起，需要使用model.stage来访问管道模型里面的iterable

标签： python machine-learning pyspark cluster-analysis k-means

【解决方案1】：

虚拟数据

from pyspark.ml.linalg import Vectors
from pyspark.ml.clustering import KMeans, KMeansModel
from pyspark.ml.pipeline import Pipeline


data = [(Vectors.dense([0.0, 0.0]),), (Vectors.dense([1.0, 1.0]),),
        (Vectors.dense([9.0, 8.0]),), (Vectors.dense([8.0, 9.0]),)]
matrix_normalized = spark.createDataFrame(data, ["scaledFeatures"])

你的代码

kmeans = KMeans() \
          .setK(3) \
          .setFeaturesCol("scaledFeatures")\
          .setPredictionCol("cluster")

# Chain indexer and tree in a Pipeline
pipeline = Pipeline(stages=[kmeans])

model = pipeline.fit(matrix_normalized)

cluster = model.transform(matrix_normalized)

只需更改最后一行

model.stages[0].clusterCenters()

[array([0.5, 0.5]), array([8., 9.]), array([9., 8.])]

【讨论】：