def test_predictOn_model(self):
"""Test that the model predicts correctly on toy data."""
stkm = StreamingKMeans()
stkm._model = StreamingKMeansModel(clusterCenters=[[1.0, 1.0],
[-1.0, 1.0],
[-1.0, -1.0],
[1.0, -1.0]],
clusterWeights=[1.0, 1.0, 1.0, 1.0])
predict_data = [[[1.5, 1.5]], [[-1.5, 1.5]], [[-1.5, -1.5]],
[[1.5, -1.5]]]
predict_data = [
self.sc.parallelize(batch, 1) for batch in predict_data
]
predict_stream = self.ssc.queueStream(predict_data)
predict_val = stkm.predictOn(predict_stream)
result = []
def update(rdd):
rdd_collect = rdd.collect()
if rdd_collect:
result.append(rdd_collect)
predict_val.foreachRDD(update)
self.ssc.start()
def condition():
self.assertEqual(result, [[0], [1], [2], [3]])
return True
eventually(condition, catch_assertions=True)
def test_predictOn_model(self):
"""Test that the model predicts correctly on toy data."""
stkm = StreamingKMeans()
stkm._model = StreamingKMeansModel(clusterCenters=[[1.0, 1.0],
[-1.0, 1.0],
[-1.0, -1.0],
[1.0, -1.0]],
clusterWeights=[1.0, 1.0, 1.0, 1.0])
predict_data = [[[1.5, 1.5]], [[-1.5, 1.5]], [[-1.5, -1.5]],
[[1.5, -1.5]]]
predict_data = [sc.parallelize(batch, 1) for batch in predict_data]
predict_stream = self.ssc.queueStream(predict_data)
predict_val = stkm.predictOn(predict_stream)
result = []
def update(rdd):
rdd_collect = rdd.collect()
if rdd_collect:
result.append(rdd_collect)
predict_val.foreachRDD(update)
t = time()
self.ssc.start()
self._ssc_wait(t, 6.0, 0.01)
self.assertEquals(result, [[0], [1], [2], [3]])
def test_streaming_kmeans():
records = get_data_from_db()
conf = SparkConf().setAppName("testingClusters").setMaster("local[2]")
sc = SparkContext(conf=conf)
sc.setLogLevel("ERROR")
initCenters = [[0.0, 0.0, 0.0], [1.0, 1.0, 1.0], [0.5, 0.5, 0.5]]
initWeights = [1.0, 1.0, 1.0]
stkm = StreamingKMeansModel(initCenters, initWeights)
recordRDD = sc.parallelize(records)
stkm = stkm.update(data=recordRDD, decayFactor=1.0, timeUnit=u"batches")
for center in stkm.centers:
print(center)
sample_data = np.array([46.5, 23.0, 1034.0]).reshape(1, -1)
record_scaled = scaler.transform(sample_data)
print("Scaled Record: ", record_scaled)
ret = stkm.predict(sc.parallelize(record_scaled))
print("******** Predicted cluster index: ")
ret.foreach(print)
for center in stkm.centers:
distance = np.linalg.norm(center - record_scaled)
similarity = 1 / (1 + distance)
print(center, distance, similarity)
stkm = stkm.update(sc.parallelize(record_scaled), 1.0, u"points")
sample_data = np.array([46.2, 23.5, 1034.32]).reshape(1, -1)
record_scaled = scaler.transform(sample_data)
print("Scaled Record: ", record_scaled)
ret = stkm.predict(sc.parallelize(record_scaled))
print("******** Predicted cluster index: ")
ret.foreach(print)
for center in stkm.centers:
distance = np.linalg.norm(center - record_scaled)
similarity = 1 / (1 + distance)
print(center, distance, similarity)
import dautil as dl
from pyspark.mllib.clustering import StreamingKMeansModel
from pyspark import SparkContext
csv_file = dl.data.get_direct_marketing_csv()
csv_rows = dl.data.read_csv(csv_file)
stkm = StreamingKMeansModel(28 * [[0., 0., 0.]], 28 * [1.])
sc = SparkContext()
for row in csv_rows:
spend = dl.data.centify(row['spend'])
if spend > 0:
history = dl.data.centify(row['history'])
data = sc.parallelize([[int(row['recency']), history, spend]])
stkm = stkm.update(data, 0., 'points')
print(stkm.centers)
import dautil as dl
from pyspark.mllib.clustering import StreamingKMeansModel
from pyspark import SparkContext
csv_file = dl.data.get_direct_marketing_csv()
csv_rows = dl.data.read_csv(csv_file)
stkm = StreamingKMeansModel(28 * [[0., 0., 0.]], 28 * [1.])
sc = SparkContext()
for row in csv_rows:
spend = dl.data.centify(row['spend'])
if spend > 0:
history = dl.data.centify(row['history'])
data = sc.parallelize([[int(row['recency']),
history, spend]])
stkm = stkm.update(data, 0., 'points')
print(stkm.centers)
if sys.argv[1]==("sset1"):
truePredictions.append(int(c[0])-1)
if sys.argv[1]==("asset1"):
truePredictions.append(int(c[0])-1)
else:
truePredictions.append(int(c[0]))
with open("datasets/"+sys.argv[1]+".txt","r") as fichero:
for linea in fichero:
points.append(linea.strip("\n").split())
for document in cursor:
centers = document["clusterCenters"]
weights = document["clusterWeights"]
stkm = StreamingKMeansModel(centers, weights)
predictions = []
for point in points:
predictions.append(stkm.predict(point))
recall = recall_score(truePredictions,predictions, average='weighted')
precision = precision_score(truePredictions, predictions, average='weighted')
f1Score = 2 * (precision * recall) / (precision + recall)
logging.info("Recall = " + str(recall) )
logging.info("Precision = " + str(precision) )
logging.info("F1-Score = " + str(f1Score))
os.environ[
'PYSPARK_SUBMIT_ARGS'] = '--packages org.apache.spark:spark-streaming-kafka-0-8_2.11:2.0.2 pyspark-shell'
sc = SparkContext(appName="PythonSparkStreamingKafka")
sc.setLogLevel("WARN")
ssc = StreamingContext(sc, 120)
initWeights = [
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0
]
offlineModel = KMeansModel.load(sc, "KMeansModel")
stkm = StreamingKMeansModel(offlineModel.clusterCenters, initWeights)
#kafkaStream = KafkaUtils.createDirectStream(ssc, ['test'], {'metadata.broker.list': 'localhost:9092'})
#lines = kafkaStream.map(lambda line: array([float(x) for x in line.split('\t')]))
consumer = KafkaConsumer('test',
bootstrap_servers=['localhost:9092'],
value_deserializer=lambda x: loads(x.decode('utf-8')),
consumer_timeout_ms=10000)
colors = [
'r', 'k', 'b', 'grey', 'darkorange', 'm', 'y', 'c', 'gold', 'slateblue',
'beige', 'coral', 'g', 'peru', 'pink'
]