def kmeans(df):
kmeans = KMeans(k=2,seed=1)
model = kmeans.fit(df)
centers = model.clusterCenters()
print len(centers)
kmFeatures = model.transform(df).select("features", "prediction")
dfwrite(kmFeatures,'kmFeatures')
def clustering(input_df, input_col_name, n):
""" KMeans and PCA """
input_df = input_df.select('state','categories','stars',input_col_name)
norm = Normalizer(inputCol=input_col_name, outputCol="features", p=1.0)
df = norm.transform(input_df)
kmeans = KMeans(k=n, seed=2)
KMmodel = kmeans.fit(df)
predicted = KMmodel.transform(df).cache()
pca = PCA(k=2, inputCol='features', outputCol="pc")
df = pca.fit(dfsample).transform(dfsample).cache()
return df
def test_kmeans_cosine_distance(self):
data = [(Vectors.dense([1.0, 1.0]),), (Vectors.dense([10.0, 10.0]),),
(Vectors.dense([1.0, 0.5]),), (Vectors.dense([10.0, 4.4]),),
(Vectors.dense([-1.0, 1.0]),), (Vectors.dense([-100.0, 90.0]),)]
df = self.spark.createDataFrame(data, ["features"])
kmeans = KMeans(k=3, seed=1, distanceMeasure="cosine")
model = kmeans.fit(df)
result = model.transform(df).collect()
self.assertTrue(result[0].prediction == result[1].prediction)
self.assertTrue(result[2].prediction == result[3].prediction)
self.assertTrue(result[4].prediction == result[5].prediction)
def elbow(elbowset, clusters):
wsseList = []
for k in clusters:
print("Training for cluster size {} ".format(k))
kmeans = KM(k = k, seed = 1)
model = kmeans.fit(elbowset)
transformed = model.transform(elbowset)
featuresAndPrediction = transformed.select("features", "prediction")
W = computeCost(featuresAndPrediction, model)
print("......................WSSE = {} ".format(W))
wsseList.append(W)
return wsseList
def test_kmeans_param(self):
algo = KMeans()
self.assertEqual(algo.getInitMode(), "k-means||")
algo.setK(10)
self.assertEqual(algo.getK(), 10)
algo.setInitSteps(10)
self.assertEqual(algo.getInitSteps(), 10)
def test_kmeans_summary(self):
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]),)]
df = self.spark.createDataFrame(data, ["features"])
kmeans = KMeans(k=2, seed=1)
model = kmeans.fit(df)
self.assertTrue(model.hasSummary)
s = model.summary
self.assertTrue(isinstance(s.predictions, DataFrame))
self.assertEqual(s.featuresCol, "features")
self.assertEqual(s.predictionCol, "prediction")
self.assertTrue(isinstance(s.cluster, DataFrame))
self.assertEqual(len(s.clusterSizes), 2)
self.assertEqual(s.k, 2)
self.assertEqual(s.numIter, 1)
def test_kmean_pmml_basic(self):
# Most of the validation is done in the Scala side, here we just check
# that we output text rather than parquet (e.g. that the format flag
# was respected).
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]),)]
df = self.spark.createDataFrame(data, ["features"])
kmeans = KMeans(k=2, seed=1)
model = kmeans.fit(df)
path = tempfile.mkdtemp()
km_path = path + "/km-pmml"
model.write().format("pmml").save(km_path)
pmml_text_list = self.sc.textFile(km_path).collect()
pmml_text = "\n".join(pmml_text_list)
self.assertIn("Apache Spark", pmml_text)
self.assertIn("PMML", pmml_text)
def kmeans(inputdir,df,alg,k):
from pyspark.ml.clustering import KMeans
from numpy import array
from math import sqrt
kmeans = KMeans(k=int(k), seed=1,initSteps=5, tol=1e-4, maxIter=20, initMode="k-means||", featuresCol="features")
model = kmeans.fit(df)
kmFeatures = model.transform(df).select("labels", "prediction")
erFeatures = model.transform(df).select("features", "prediction")
###Evaluation
rows = erFeatures.collect()
WSSSE = 0
for i in rows:
WSSSE += sqrt(sum([x**2 for x in (model.clusterCenters()[i[1]]-i[0])]))
print("Within Set Sum of Squared Error = " + str(WSSSE))
output_data = writeOutClu(inputdir,kmFeatures,alg,k,WSSSE)
return output_data
def doKMeans(data):
# groupedData = data.groupBy("station_id","stationIndex").agg(avg("max_temp"),avg("med_temp"),avg("min_temp"),avg("max_pressure"),avg("min_pressure"),avg("precip"),avg("insolation"))
# selectedData = groupedData.select("station_id","stationIndex","avg(max_temp)","avg(med_temp)","avg(min_temp)","avg(max_pressure)","avg(min_pressure)","avg(precip)","avg(insolation)")
# columnsToPredict = ["stationIndex","avg(max_temp)","avg(med_temp)","avg(min_temp)","avg(max_pressure)","avg(min_pressure)","avg(precip)","avg(insolation)"]
# assembler = VectorAssembler(inputCols=columnsToPredict,outputCol="features")
# assembledData = assembler.transform(selectedData)
# groupedData = data.groupBy("station_id","stationIndex").agg(avg("max_temp"),avg("med_temp"),avg("min_temp"),avg("max_pressure"),avg("min_pressure"),avg("precip"),avg("insolation"))
# selectedData = groupedData.select("station_id","stationIndex","avg(max_temp)","avg(med_temp)","avg(min_temp)","avg(max_pressure)","avg(min_pressure)","avg(precip)","avg(insolation)")
columnsToPredict = ["stationIndex", "max_temp", "med_temp", "min_temp", "max_pressure", "min_pressure", "precip",
"insolation"]
assembler = VectorAssembler(inputCols=columnsToPredict, outputCol="features")
assembledData = assembler.transform(data)
sampledData = assembledData
feature_data = sampledData.withColumn("label", sampledData.stationIndex).withColumn("features",
sampledData.features)
print("Sampling...")
test_data = feature_data.sample(False, 0.1)
train_data = feature_data.sample(False, 0.9)
print("Test data: " + str(test_data.count()) + " , Train data: " + str(train_data.count()))
# Trains a k-means model.
# kmeans = KMeans().setK(stations.count()).setSeed(1)
kmeans = KMeans().setK(stations.count()).setSeed(1)
model = kmeans.fit(feature_data)
# Evaluate clustering by computing Within Set Sum of Squared Errors.
wssse = model.computeCost(feature_data)
print("Within Set Sum of Squared Errors = " + str(wssse))
# Shows the result.
centers = model.clusterCenters()
print("Cluster Centers: ")
for center in centers:
print(center)
print("Predicting...")
predictions = model.transform(test_data)
# predictions.show()
predictions.select("station_id", "stationIndex", "label", "prediction").show()
def _fit(self, dataset):
k = self.getK()
vocabulary = self.getVocabulary()
# Trains a k-means model on word-vectors
kmeans = KMeans(featuresCol="vector", predictionCol="cluster", initMode="random") \
.setK(k).setSeed(1)
model_kmeans = kmeans.fit(vocabulary)
data_vocabulary = model_kmeans.transform(vocabulary)
# Create and broadcast the dictionary <word, cluster>
rdd_dictionary = data_vocabulary.select('word','cluster') \
.map(lambda row: (row['word'],str(row['cluster'])))
dictionary = dict(rdd_dictionary.collect())
return (WordClusterModel().setInputCol(
self.getInputCol()).setPredictionCol(
self.getPredictionCol()).setDictionary(dictionary).setK(k))
def train(df, hiperparameter):
'''
KMeans training, returning KMeans model.
input: - Dataframe
- config (configurasi hiperparameter)
return: kmeans model
'''
k_means = KMeans(featuresCol=hiperparameter['featuresCol'],
predictionCol=hiperparameter['predictionCol'],
k=hiperparameter['k'],
initMode=hiperparameter['initMode'],
initSteps=hiperparameter['initSteps'],
tol=hiperparameter['tol'],
maxIter=hiperparameter['maxIter'],
seed=hiperparameter['seed'])
model = k_means.fit(df)
return model
def test_kmeans(self):
kmeans = KMeans(k=2, seed=1)
path = tempfile.mkdtemp()
km_path = path + "/km"
kmeans.save(km_path)
kmeans2 = KMeans.load(km_path)
self.assertEqual(kmeans.uid, kmeans2.uid)
self.assertEqual(type(kmeans.uid), type(kmeans2.uid))
self.assertEqual(kmeans2.uid, kmeans2.k.parent,
"Loaded KMeans instance uid (%s) did not match Param's uid (%s)"
% (kmeans2.uid, kmeans2.k.parent))
self.assertEqual(kmeans._defaultParamMap[kmeans.k], kmeans2._defaultParamMap[kmeans2.k],
"Loaded KMeans instance default params did not match " +
"original defaults")
try:
rmtree(path)
except OSError:
pass
def test_kmeans(self):
kmeans = KMeans(k=2, seed=1)
path = tempfile.mkdtemp()
km_path = path + "/km"
kmeans.save(km_path)
kmeans2 = KMeans.load(km_path)
self.assertEqual(kmeans.uid, kmeans2.uid)
self.assertEqual(type(kmeans.uid), type(kmeans2.uid))
self.assertEqual(kmeans2.uid, kmeans2.k.parent,
"Loaded KMeans instance uid (%s) did not match Param's uid (%s)"
% (kmeans2.uid, kmeans2.k.parent))
self.assertEqual(kmeans._defaultParamMap[kmeans.k], kmeans2._defaultParamMap[kmeans2.k],
"Loaded KMeans instance default params did not match " +
"original defaults")
try:
rmtree(path)
except OSError:
pass
def train(model="kmean", scaledData=None, test_id=4):
"""
train models based on the model type
args:
model: str, the model type
scaledData: the scaled data from the output of norm_data function
"""
for i in range(5):
kmeans = KMeans().setK(2).setSeed(1)
model = kmeans.fit(scaledData)
predictions = model.transform(scaledData)
evaluator = ClusteringEvaluator()
silhouette = evaluator.evaluate(predictions)
centers = model.clusterCenters()
for center in centers:
print(center)
predictions.rdd.saveAsPickleFile("k_mean_prediction_worker2_testid_" +
str(test_id) + ".pkl")
def _compute_cluster_analysis(spark_df, clusters=5):
numeric_columns = list(map(lambda col_dtype: col_dtype[0], spark_df.dtypes))
if (len(numeric_columns) == 0):
raise ValueError("The provided spark dataframe does not contain any numeric columns. "
"Cannot compute cluster analysis with k-means on categorical columns. "
"The numeric datatypes are: {}" \
" and the number of numeric datatypes in the dataframe is: {} ({})".format(
constants.SPARK_CONFIG.SPARK_NUMERIC_TYPES, len(spark_df.dtypes), spark_df.dtypes))
if (len(numeric_columns) == 1):
raise ValueError("The provided spark dataframe does contains only one numeric column. "
"Cluster analysis will filter out numeric columns and then "
"use pca to reduce dataset dimension to 2 dimensions and "
"then apply KMeans, this is not possible when the input data have only one numeric column."
"The numeric datatypes are: {}"
" and the number of numeric datatypes in the dataframe is: {} ({})".format(
constants.SPARK_CONFIG.SPARK_NUMERIC_TYPES, len(spark_df.dtypes), spark_df.dtypes))
vecAssembler = VectorAssembler(inputCols=numeric_columns,
outputCol=constants.FEATURE_STORE.CLUSTERING_ANALYSIS_INPUT_COLUMN)
spark_df_1 = vecAssembler.transform(spark_df)
kmeans = KMeans(k=clusters, seed=1, maxIter=20,
featuresCol=constants.FEATURE_STORE.CLUSTERING_ANALYSIS_INPUT_COLUMN,
predictionCol=constants.FEATURE_STORE.CLUSTERING_ANALYSIS_OUTPUT_COLUMN)
model = kmeans.fit(spark_df_1.select(constants.FEATURE_STORE.CLUSTERING_ANALYSIS_INPUT_COLUMN))
spark_df_2 = model.transform(spark_df_1)
spark_df_3 = spark_df_2.select([constants.FEATURE_STORE.CLUSTERING_ANALYSIS_INPUT_COLUMN,
constants.FEATURE_STORE.CLUSTERING_ANALYSIS_OUTPUT_COLUMN])
count = spark_df_3.count()
if count < constants.FEATURE_STORE.CLUSTERING_ANALYSIS_SAMPLE_SIZE:
spark_df_4 = spark_df_3
else:
spark_df_4 = spark_df_3.sample(True,
float(constants.FEATURE_STORE.CLUSTERING_ANALYSIS_SAMPLE_SIZE) / float(count))
pca = PCA(k=2,
inputCol=constants.FEATURE_STORE.CLUSTERING_ANALYSIS_INPUT_COLUMN,
outputCol=constants.FEATURE_STORE.CLUSTERING_ANALYSIS_PCA_COLUMN)
model = pca.fit(spark_df_4)
spark_df_5 = model.transform(spark_df_4).select([constants.FEATURE_STORE.CLUSTERING_ANALYSIS_PCA_COLUMN,
constants.FEATURE_STORE.CLUSTERING_ANALYSIS_OUTPUT_COLUMN])
spark_df_6 = spark_df_5.withColumnRenamed(
constants.FEATURE_STORE.CLUSTERING_ANALYSIS_PCA_COLUMN,
constants.FEATURE_STORE.CLUSTERING_ANALYSIS_FEATURES_COLUMN)
spark_df_7 = spark_df_6.withColumnRenamed(constants.FEATURE_STORE.CLUSTERING_ANALYSIS_OUTPUT_COLUMN, "clusters")
return json.loads(spark_df_7.toPandas().to_json())
def create_clusters(sqlContext):
specific_data = sqlContext.sql(
'SELECT Description, Date, Latitude, Longitude FROM crime_data')
#where `Primary Type` == "HOMOCIDE" or `Primary Type` == "ASSAULT"
specific_data.filter(specific_data.Description != "SIMPLE")
specific_data = specific_data.rdd.filter(lambda x: datetime.strptime(
x[1], "%m/%d/%Y %I:%M:%S %p").weekday() in [5, 6]).toDF()
specific_data = specific_data.sample(False, .0001)
#create different k-means outputs depending on the date since that is what we are testing
vecAssembler = VectorAssembler(inputCols=['Latitude', 'Longitude'],
outputCol="features")
df_kmeans = vecAssembler.transform(specific_data).select(
'Description', 'features', 'Latitude', 'Longitude')
# df_kmeans.show()
# cost = [0] * 40
# for k in range(41,50):
# kmeans = KMeans().setK(k).setSeed(1).setFeaturesCol("features")
# model = kmeans.fit(df_kmeans.sample(False,0.1, seed=42))
# cost[k-41] = model.computeCost(df_kmeans) # requires Spark 2.0 or later
# print (cost)
# 48 for k since this is where it starts to have diminishing returns
k = 48
kmeans = KMeans().setK(k).setSeed(1).setFeaturesCol("features")
model = kmeans.fit(df_kmeans)
centers = model.clusterCenters()
# print("Cluster Centers: ")
# for center in centers:
# print(center)
#TODO: arrange based on season, time of day, etc
transformed = model.transform(df_kmeans).select('Description',
'prediction', 'Latitude',
'Longitude')
transformed.show()
#return_rdd = transformed.rdd.map(lambda x: (x.prediction, (x.Description, (x.Latitude, x.Longitude))))
create_visuals(transformed)
def clustering(poses):
print(len(poses))
dff = map(lambda x: (0,(Vectors.dense(x[0:2]))), poses)
mydf = spark.createDataFrame(dff,schema=["None","features"])
k = 2
cost = 99999999999
while(cost >= 20):
kmeans = KMeans().setK(k).setSeed(1)
model = kmeans.fit(mydf)
cost = model.computeCost(mydf)/len(poses)
k+=1
print("K: ", k, " Cost: ", cost)
centers = model.clusterCenters()
centroids = []
for center in centers:
centroids.append(center)
prediction = model.transform(mydf).select('prediction').collect()
labels = [p.prediction for p in prediction ]
return centroids, labels
def getSilhouette(df, model='KMeans'):
silhouette_ls = []
if model == 'KMeans':
for i in range(1, 10):
kmeans = KMeans().setK(i + 1).setSeed(123)
model_k = kmeans.fit(df)
# Make predictions
predictions = model_k.transform(df)
# Evaluate clustering by computing Silhouette score
evaluator = ClusteringEvaluator()
silhouette_ls.append(round(evaluator.evaluate(predictions), 2))
best_k = silhouette_ls.index(np.max(silhouette_ls)) + 2
print(
'****************************SILHOUETTE*************************************************'
)
print(
f'The best K is: {best_k} associated with a silhoutte of: {np.max(silhouette_ls)}'
)
return silhouette_ls, best_k
def test_kmeans_summary(self):
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]), ),
]
df = self.spark.createDataFrame(data, ["features"])
kmeans = KMeans(k=2, seed=1)
model = kmeans.fit(df)
self.assertTrue(model.hasSummary)
s = model.summary
self.assertTrue(isinstance(s.predictions, DataFrame))
self.assertEqual(s.featuresCol, "features")
self.assertEqual(s.predictionCol, "prediction")
self.assertTrue(isinstance(s.cluster, DataFrame))
self.assertEqual(len(s.clusterSizes), 2)
self.assertEqual(s.k, 2)
self.assertEqual(s.numIter, 1)
def main():
cfg = Path(__file__).parent.joinpath("kmeans_config.yaml")
cfg = load_config(str(cfg))
spark = init_spark(cfg)
kmeans = KMeans(k=3, seed=0)
X = read(cfg, spark)
fit(cfg, kmeans, X)
print(timings)
def create_kmeans_pipeline_from(data_frame,
feature_extractor,
total_clusters,
max_iterations):
"""
Creates a pipeline for tokenizing the document to words,
removing stop words, and add TF-IDF, and finally does
clustering using k-means algorithm
"""
tokenizer_transformer = RegexTokenizer(inputCol="doc_text",
outputCol="words",
pattern="\\W")
stop_words_transformer = StopWordsRemover(inputCol="words",
outputCol="filtered_words")
pipeline_stages = [tokenizer_transformer, stop_words_transformer,]
if feature_extractor == 'TFIDF':
# create TF counter using CountVectorizer
tf_estimator = CountVectorizer(inputCol="filtered_words", outputCol="TF")
# create inverse-document-frequency counter
idf_estimator = IDF(inputCol="TF", outputCol="features")
# add them to the pipeline stages
pipeline_stages.extend([tf_estimator, idf_estimator])
elif feature_extractor == 'Word2Vec':
# create word2vec feature extractor
w2v_estimator = Word2Vec(inputCol="filtered_words", outputCol="features")
# add this to pipeline stage
pipeline_stages.append(w2v_estimator)
else:
raise ValueError('Unknown feature extractor:' % feature_extractor)
# create KMeans clustering
kmeans = KMeans(k=total_clusters,
featuresCol="features",
predictionCol="DocumentClass",
seed=1,
maxIter=max_iterations)
# finally add Kmeans to the pipeline
# which takes "features" output from the previous stage and
# does the prediction.
# NOTE:
# For document clustering cosine_similarity measure is the preferred one.
# This pipeline uses SSE method
#
pipeline_stages.append(kmeans)
return Pipeline(stages=pipeline_stages).fit(data_frame)
def kmeans(df, optimalK, seed=1):
"""
Input:
df: pyspark dataframe
optimalK: int, k for cluster
seed: int, random seed
Output:
new_df_kmeans_transformed: pyspark dataframe, with the cluster prediction
model: k-means model
"""
# transfer to "features" for pyspark ML
scaled_column = [col for col in df.columns if "scaled" in col]
df_vec = get_feature(df, scaled_column)
kmeans = KMeans(k=optimalK, seed=seed)
model = kmeans.fit(df_vec)
new_df_kmeans_transformed = model.transform(df_vec)
return new_df_kmeans_transformed, model
def test_kmean_pmml_basic(self):
# Most of the validation is done in the Scala side, here we just check
# that we output text rather than parquet (e.g. that the format flag
# was respected).
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]), ),
]
df = self.spark.createDataFrame(data, ["features"])
kmeans = KMeans(k=2, seed=1)
model = kmeans.fit(df)
path = tempfile.mkdtemp()
km_path = path + "/km-pmml"
model.write().format("pmml").save(km_path)
pmml_text_list = self.sc.textFile(km_path).collect()
pmml_text = "\n".join(pmml_text_list)
self.assertIn("Apache Spark", pmml_text)
self.assertIn("PMML", pmml_text)
def kmeans():
spark = SparkSession \
.builder \
.appName('my_first_app_name') \
.getOrCreate()
df = spark.read.csv('./seeds_dataset.csv', header=True, inferSchema=True)
df.show(5)
assembler = VectorAssembler(inputCols=df.columns, outputCol='features')
final_df = assembler.transform(df)
final_df.show(3)
final_df.take(1)
# instantiated kmeans with 3 cluster
kmeans = KMeans(k=3)
# fitting the model
model = kmeans.fit(final_df)
centers = model.clusterCenters()
print(centers)
model.transform(final_df).select('prediction').show()
def cluster():
ld = load(open(DATAP + '\\temp\olangdict.json', 'r', encoding='UTF-8'))
spark = SparkSession.builder\
.master("local")\
.appName("Word Count")\
.config("spark.some.config.option", "some-value")\
.getOrCreate()
df = spark.createDataFrame([["0"], ["1"], ["2"], ["3"], ["4"]], ["id"])
df.show()
vecAssembler = VectorAssembler(inputCols=["feat1", "feat2"],
outputCol="features")
new_df = vecAssembler.transform(df)
kmeans = KMeans(k=2, seed=1) # 2 clusters here
model = kmeans.fit(new_df.select('features'))
transformed = model.transform(new_df)
print(transformed.show())
def analysing_emissions_data(spark, co2_emisssion_data):
# creating feature vector for sending as input to ML models
vecAssembler = VectorAssembler(inputCols=['change_in_emissions_scaled'],
outputCol="features")
# adding feature vector to our aperk dataframe
co2_emisssion_data = vecAssembler.setHandleInvalid("skip").transform(
co2_emisssion_data)
# creating Kmeans object (7 clusters)
kmeans = KMeans(k=7)
# clustering operation
model = kmeans.fit(co2_emisssion_data.select('features'))
# adding column of predicted clusters to our dataframe
co2_emisssion_data = model.transform(co2_emisssion_data)
return co2_emisssion_data.drop("features")
def clustering(fp, cols):
assembler = VectorAssembler(inputCols=cols, outputCol='features')
assembled_data = assembler.transform(fp)
scaler = StandardScaler(inputCol='features', outputCol='scaledFeatures')
print('dkks' scaler)
#scaler.show()
scaler_model = scaler.fit(assembled_data)
scaled_data = scaler_model.transform(assembled_data)
#scaled_data.printSchema()
#scaled_data.show(4)
#scaled_data.select('scaledFeatures').show()
k_means_3 = KMeans(featuresCol='scaledFeatures', k=10) #clusters = KMeans(fp, 5, maxIterations=10, initializationMode="random")#clusters.show()
model_k3 = k_means_3.fit(scaled_data)
model_k3_data = model_k3.transform(scaled_data)
#details = model_k3_data
#model_k3_data.groupBy('prediction').count().show()
#details.show(50)
return model_k3_data
def __k_mean(cls, df, k_clusters, xnorm, label_name):
from sklearn.cluster import KMeans
kmeans_cat = label_name
if xnorm.shape[1] > 0:
#n_clusters = elbow point
start_time = time.time()
estimator = KMeans(n_clusters=k_clusters)
estimator.fit(xnorm)
y_pred = estimator.predict(xnorm)
res=estimator.__dict__
#print(res['cluster_centers_'])
cluster_center = res['cluster_centers_']
print("training time: ", time.time()-start_time, "(sec)")
else:
#no split
y_pred = 0
cluster_center = []
print("no kmeans split")
df[kmeans_cat] = y_pred + 1
return df, kmeans_cat, cluster_center
def buildAnomalyDetector(data, normalizeFunction):
normalizedData = data.map(normalizeFunction)
normalizedData.cache()
model = KMeans.train(normalizedData, 150, maxIterations=10, epsilon=1.0e-6)
normalizedData.unpersist()
distances = normalizedData.map(lambda datum: distToCentroid(datum, model))
threshold = distances.top(100).pop()
def f(datum):
return distToCentroid(normalizeFunction(datum), model) > threshold
return f
def executeKmeans(k, username, jobNum, input_file_name, model_file_name,
prediction_file_name, features_col):
file_path_in_hdfs = username + '/' + jobNum + '/'
ip_file_in_hdfs = file_path_in_hdfs + input_file_name
df_kmeans = preprocessData(ip_file_in_hdfs, features_col)
kmeans = KMeans().setK(k).setSeed(1).setFeaturesCol("features")
model = kmeans.fit(df_kmeans)
centers = model.clusterCenters()
model_file_in_hdfs = file_path_in_hdfs + model_file_name
saveModelinHDFS.apply_async(
(file_path_in_hdfs, json.dumps(centers,
cls=NumpyEncoder), model_file_in_hdfs))
predictions = makePredictions(df_kmeans, model)
prediction_file_in_hdfs = file_path_in_hdfs + prediction_file_name
savePredictioninHDFS.apply_async(
(file_path_in_hdfs,
json.dumps(predictions.values.tolist(),
cls=NumpyEncoder), prediction_file_in_hdfs))
def main(argv):
#se instancia el contexto de spark.
sc = SparkContext(appName="KMeans-Clustering-dhoyoso-dsernae")
#se inicia sesion en spark.
spark = SparkSession(sc)
#se guarda el lenguaje a partir del cual se quitaran las stop words.
language = argv[4] #"spanish"
#se guarda la ruta para la salida de los clusters.
pathout = argv[3]
#se guarda la ruta de la cual se leeran los archivos.
path = argv[2] #"hdfs:///user/dhoyoso/datasets/dataset/"
#se guarda el numero de clusters que se desea hacer.
k = int(argv[1]) #4
#se sacan los archivos a procesar a partir de la ruta.
files = sc.wholeTextFiles(path)
#se crea la estructura del dataframe; 2 columnas una para la ruta y otra para el texto.
schema = StructType([
StructField("path", StringType(), True),
StructField("text", StringType(), True)
])
#se crea el dataframe a partir de la estructura y los archivos.
df = spark.createDataFrame(files, schema)
#se tokeniza el texto usando la clase de Ml tokenizer.
tokenizer = Tokenizer(inputCol="text", outputCol="tokens")
#se le dice al stop words remover que idioma es el que estamos tratando.
StopWordsRemover.loadDefaultStopWords(language)
#se remueven las stopwords de los tokens.
stopWords = StopWordsRemover(inputCol="tokens",
outputCol="stopWordsRemovedTokens")
#se hace el hashing tf de los tokens restantes.
hashingTF = HashingTF(inputCol="stopWordsRemovedTokens",
outputCol="rawFeatures",
numFeatures=2000)
#se hace el idf de la salida del hashingTF
idf = IDF(inputCol="rawFeatures", outputCol="features", minDocFreq=1)
#se inicializa el kmeans con el idf y el k deseado.
kmeans = KMeans(k=k)
#creacion del mapa de transformaciones.
pipeline = Pipeline(stages=[tokenizer, stopWords, hashingTF, idf, kmeans])
#inserta el dataframe como el inicio de las transformaciones
model = pipeline.fit(df)
#ejecuta las trasformaciones mapeadas y guarda el resultado
results = model.transform(df)
results.cache()
#se corta la ruta para dejar solo el nombre y su respectivo cluster(prediction).
split_col = split(results['path'], '/')
results = results.withColumn('docname', split_col.getItem(7))
df = results.select("docname", "prediction")
#se agrupan los documentos del mismo cluster en cluster_docs_list y se guardan en el path de salida como un json.
grouped = df.groupBy(['prediction']).agg(
collect_list("docname").alias('cluster_docs_list'))
grouped.coalesce(1).write.json(path=pathout, mode="overwrite")
def cluster(dataframe, k, maxIter=50):
clf = KMeans(k=k, maxIter=maxIter)
model = clf.fit(dataframe)
print('=================================')
print('Cost:', model.summary.trainingCost)
# print('Cluster:', model.summary.cluster)
# print('Cluster centers:', model.clusterCenters())
print('Cluster size:', model.summary.clusterSizes)
print('Iter:', model.summary.numIter)
# print('Predictions:', model.summary.predictions)
print('k:', model.summary.k)
print('featuresCol:', model.summary.featuresCol)
print('predictionCol:', model.summary.predictionCol)
# print(dir(model.summary))
# print('=================================')
# print('Saving centers...')
# np.save('tmp/centers' + str(options.k), model.clusterCenters())
# print('Saving predictions...')
# model.summary.predictions.write.format('parquet').saveAsTable('predictions' + str(options.k))
return model.clusterCenters()
def kmeans_pipeline(self, inputCols, cluster_count=None, max_cluster=None):
if max_cluster != None:
self._max_cluster = max_cluster
assembler = VectorAssembler(inputCols=inputCols, outputCol="features")
assembled = assembler.transform(self._data_frame)
mmScaler = StandardScaler(inputCol="features",
outputCol="featuresCol",
withStd=True,
withMean=False)
scale_model = mmScaler.fit(assembled)
vectorized_data = scale_model.transform(assembled)
if cluster_count == None:
cluster_count_array = list(range(2, self._max_cluster))
wssse_output = []
for n_cluster in cluster_count_array:
kmeans = KMeans().setK(n_cluster).setSeed(1)
kmeans_model = kmeans.fit(vectorized_data)
wssse = kmeans_model.computeCost(vectorized_data)
wssse_output.append(wssse)
wssse_dict = dict(list(zip(cluster_count_array, wssse_output)))
cluster_count = min(wssse_dict, key=wssse_dict.get)
kmeans = KMeans().setK(cluster_count).setSeed(1)
kmeans_model = kmeans.fit(vectorized_data)
wssse = kmeans_model.computeCost(vectorized_data)
centers = kmeans_model.clusterCenters()
cluster_prediction = kmeans_model.transform(vectorized_data)
else:
wssse_dict = {}
kmeans = KMeans().setK(cluster_count).setSeed(1)
kmeans_model = kmeans.fit(vectorized_data)
wssse = kmeans_model.computeCost(vectorized_data)
centers = kmeans_model.clusterCenters()
cluster_prediction = kmeans_model.transform(vectorized_data)
self._kmeans_result["cluster_count"] = cluster_count
self._kmeans_result["wssse"] = wssse
self._kmeans_result["wssse_dict"] = wssse_dict
self._kmeans_result["centers"] = centers
self._kmeans_result["cluster_count"] = cluster_count
self._kmeans_result["inputCols"] = inputCols
# cluster_prediction = cluster_prediction.withColumn("prediction", cluster_prediction["prediction"].cast(StringType()))
# print cluster_prediction.printSchema()
self._predictedData = cluster_prediction
def newTrain():
fileSave = "/home/hadoop/data_school/sparkMlib/KMeans"
# 男: 1, 女: 2
df = spark.read.format('csv').option('header', 'true').load(fileSave).fillna('0')
df = df.where(df.TotalFee != '0').where(df.DiseaseCode == '13104')
df = df.withColumn("Sex", df.Sex.cast(IntegerType())) \
.withColumn("Age", df.Age.cast(IntegerType())) \
.withColumn("TotalFee", df.TotalFee.cast(FloatType()))
# vecAss = VectorAssembler(inputCols=df.columns[2:], outputCol='feature')
# data = vecAss.transform(df).select("feature")
# data.show()
data = df.drop("Sex", "DiseaseCode")
data.show()
# 转换数据
featureCreator = VectorAssembler(inputCols=data.columns, outputCol='features')
data = featureCreator.transform(data)
distance = []
for k in range(2, 10):
# 评估器
kmeans = KMeans(featuresCol='features').setK(k)
# 模型拟合
model = kmeans.fit(data)
# 聚合
test = model.transform(data).select('features', 'prediction')
evaluator = ClusteringEvaluator()
evaResult = evaluator.evaluate(test)
print("the distance = " + str(evaResult))
distance.append(evaResult)
import matplotlib.pyplot as plt
plt.figure()
x = [i for i in range(2, 10)]
plt.plot(x, distance)
plt.xlabel("K")
plt.ylabel("Distance")
plt.show()
def topic_generator(subreddit_input):
subreddit_filter = requests.get(
url + 'reddit_post.json?orderBy="subreddit"&equalTo="' +
str(subreddit_input) + '"')
subreddits = json.loads(subreddit_filter.text)
results = []
for x in subreddits:
try:
results.append(subreddits[x])
except KeyError:
continue
data = pd.DataFrame.from_dict(results, orient='columns')
data1 = spark.createDataFrame(pd.DataFrame(data["title"]))
#text clean
clean_data_udf = udf(clean_data, StringType())
data1 = data1.withColumn("new_title", clean_data_udf("title"))
#text tokenizer
tokenizer = Tokenizer(inputCol="new_title", outputCol="words")
data1 = tokenizer.transform(data1)
#stopwords removal
remover = StopWordsRemover(inputCol="words", outputCol="rm_words")
data1 = remover.transform(data1)
#TFIDF vectorization
hashingTF = HashingTF(inputCol="rm_words",
outputCol="rawFeatures",
numFeatures=2000)
data1 = hashingTF.transform(data1)
#Document frequency
idf = IDF(inputCol="rawFeatures", outputCol="features")
idfModel = idf.fit(data1)
data1 = idfModel.transform(data1)
#Kmeans topic clustering
kmeans = KMeans(k=2, featuresCol="features").setSeed(1)
kmeans_model = kmeans.fit(data1)
data1 = kmeans_model.transform(data1)
data["prediction"] = data1.select("prediction").toPandas()
return data
def kmeans_info(i, df, inputCols):
vecAssembler = VectorAssembler(inputCols=inputCols, outputCol='features')
df = vecAssembler.transform(df)
info = '---> With ' + str(i) + ' clusters \n'
kmeans = KMeans(k=i, seed=1)
model = kmeans.fit(df.select('features'))
# Make predictions
predictions = model.transform(df)
# Evaluate clustering by computing Silhouette score
evaluator = ClusteringEvaluator()
silhouette = evaluator.evaluate(predictions)
info += "Silhouette with squared euclidean distance = " + str(
silhouette) + '\n'
info += "Cluster Centers: " + '\n'
ctr = []
centers = model.clusterCenters()
for center in centers:
ctr.append(center)
info += str(center) + '\n'
info += '--------------------\n'
return info
def __process_dataset(self):
"""Making model from dataset
"""
data = []
for i in range(len(self.model)):
logger.info("Training consumers batch {}".format(i))
logger.info("Assembling vector!")
assembler = VectorAssembler(inputCols=["_c12"],
outputCol='features')
data.append(assembler.transform(self.model[i]))
logger.info("Training Model!")
kmeans = KMeans().setK(10).setSeed(1)
self.model[i] = kmeans.fit(data[i])
logger.info("Making Preditction!")
self.predictions = self.model[i].transform(data[i])
logger.info("Clustering model built!")
def visualizationInR(rawData):
def preprocessing(line):
values = line.split(",")
del values[1:4]
values.pop()
return Vectors.dense(map(lambda x: float(x), values))
data = rawData.map(preprocessing).cache()
model = KMeans.train(data, 100, maxIterations=10, epsilon=1.0e-6)
sample = data.map(lambda datum: model.predict(datum) + "," + ",".join(datum)).sample(False, fraction=0.05, seed=None)
sample.saveAsTextFile("file:///user/ds/sample")
def clustering(player_profile):
vecAssembler = VectorAssembler(
inputCols=["fouls", "goals", "owngoals", "pass_acc", "shots", "matches"], outputCol="features")
print(player_profile.printSchema())
player_profile = player_profile.drop("name")
new_df = vecAssembler.transform(player_profile)
kmeans = KMeans(k=5)
model = kmeans.fit(new_df.select("features"))
# Make predictions
predictions = model.transform(new_df)
predictions.show()
# Evaluate clustering by computing Silhouette score
evaluator = ClusteringEvaluator()
silhouette = evaluator.evaluate(predictions)
print("Silhouette with squared euclidean distance = " + str(silhouette))
# Shows the result.
centers = model.clusterCenters()
print("Cluster Centers: ")
for center in centers:
print(center)
return predictions, centers
def cluster():
ld = load(open(DATAP+'\\temp\olangdict.json','r',encoding='UTF-8'))
spark = SparkSession.builder\
.master("local")\
.appName("Word Count")\
.config("spark.some.config.option", "some-value")\
.getOrCreate()
df = spark.createDataFrame([["0"],
["1"],
["2"],
["3"],
["4"]],
["id"])
df.show()
vecAssembler = VectorAssembler(inputCols=["feat1", "feat2"], outputCol="features")
new_df = vecAssembler.transform(df)
kmeans = KMeans(k=2, seed=1) # 2 clusters here
model = kmeans.fit(new_df.select('features'))
transformed = model.transform(new_df)
print(transformed.show())
trainingData = VectorAssembler(inputCols=["duration", "tempo", "loudness"], outputCol="features").transform(
table("songsTable")
)
# COMMAND ----------
# MAGIC %md We can now pass this new DataFrame to the `KMeans` model and ask it to categorize different rows in our data to two different classes (`setK(2)`). We place the model in a variable named `model`.
# MAGIC
# MAGIC **Note:** This command multiple spark jobs (one job per iteration in the KMeans algorithm). You will see the progress bar starting over and over again.
# COMMAND ----------
from pyspark.ml.clustering import KMeans
model = KMeans().setK(2).fit(trainingData)
# COMMAND ----------
# MAGIC %md To see the result of our clustering, we produce a scatter plot matrix that shows interaction between input variables and learned clusters. To get that we apply the model on the original data and pick four columns: `prediction` and the original features (`duration`, `tempo`, and `loudness`).
# COMMAND ----------
transformed = model.transform(trainingData).select("duration", "tempo", "loudness", "prediction")
# COMMAND ----------
# MAGIC %md To comfortably visualize the data we produce a random sample.
# MAGIC Remember the `display()` function? We can use it to produce a nicely rendered table of transformed DataFrame.
# COMMAND ----------
# COMMAND ---------- fittedPipeline = transformationPipeline.fit(trainDataFrame) # COMMAND ---------- transformedTraining = fittedPipeline.transform(trainDataFrame) # COMMAND ---------- from pyspark.ml.clustering import KMeans kmeans = KMeans()\ .setK(20)\ .setSeed(1L) # COMMAND ---------- kmModel = kmeans.fit(transformedTraining) # COMMAND ---------- transformedTest = fittedPipeline.transform(testDataFrame) # COMMAND ----------
from pyspark.mllib.linalg import Vectors
from pyspark.ml.clustering import KMeans
from pyspark import SparkContext
from pyspark.sql import SQLContext
# sc = SparkContext(appName="test")
# sqlContext = SQLContext(sc)
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]),)]
df = sqlContext.createDataFrame(data, ["features"])
kmeans = KMeans(k=2, seed=1)
model = kmeans.fit(df)
centers = model.clusterCenters()
model.transform(df).select("features", "prediction").collect()
# For now, analysis is still required. We cache the output because we are going to perform
# multiple runs on the dataset.
df0 = tfs.analyze(df).cache()
mllib_df.count()
df0.count()
np.random.seed(2)
init_centers = np.random.randn(k, num_features)
start_centers = init_centers
dataframe = df0
ta_0 = time.time()
kmeans = KMeans().setK(k).setSeed(1).setFeaturesCol(FEATURES_COL).setInitMode(
"random").setMaxIter(num_iters)
mod = kmeans.fit(mllib_df)
ta_1 = time.time()
tb_0 = time.time()
(centers, agg_distances) = kmeanstf(df0, init_centers, num_iters=num_iters, tf_aggregate=False)
tb_1 = time.time()
tc_0 = time.time()
(centers, agg_distances) = kmeanstf(df0, init_centers, num_iters=num_iters, tf_aggregate=True)
tc_1 = time.time()
mllib_dt = ta_1 - ta_0
tf_dt = tb_1 - tb_0
tf2_dt = tc_1 - tc_0
sales = va.transform(spark.read.format("csv")
.option("header", "true")
.option("inferSchema", "true")
.load("/data/retail-data/by-day/*.csv")
.limit(50)
.coalesce(1)
.where("Description IS NOT NULL"))
sales.cache()
# COMMAND ----------
from pyspark.ml.clustering import KMeans
km = KMeans().setK(5)
print km.explainParams()
kmModel = km.fit(sales)
# COMMAND ----------
summary = kmModel.summary
print summary.clusterSizes # number of points
kmModel.computeCost(sales)
centers = kmModel.clusterCenters()
print("Cluster Centers: ")
for center in centers:
print(center)
def assign_cluster(data):
"""Train kmeans on rescaled data and then label the rescaled data."""
kmeans = KMeans(k=2, seed=1, featuresCol="features_scaled", predictionCol="label")
model = kmeans.fit(data)
label_df = model.transform(data)
return label_df
# COMMAND ----------
display(transformed)
# COMMAND ----------
# MAGIC %md
# MAGIC #### K-Means Visualized
# COMMAND ----------
modelCenters = []
iterations = [0, 2, 4, 7, 10, 20]
for i in iterations:
kmeans = KMeans(k=3, seed=5, maxIter=i, initSteps=1)
model = kmeans.fit(irisTwoFeatures)
modelCenters.append(model.clusterCenters())
# COMMAND ----------
print 'modelCenters:'
for centroids in modelCenters:
print centroids
# COMMAND ----------
import matplotlib.pyplot as plt
import matplotlib.cm as cm
import numpy as np
twDF = tw.map(lambda p: Row(text=p)).toDF()
t0 = time.time()
word2Vec = Word2Vec(vectorSize=100, minCount=5, stepSize=0.025, inputCol="text", outputCol="result")
modelW2V = word2Vec.fit(twDF)
wordVectorsDF = modelW2V.getVectors()
timeW2V = time.time() - t0
## Train K-means on top of the Word2Vec matrix:
t0 = time.time()
vocabSize = wordVectorsDF.count()
K = int(math.floor(math.sqrt(float(vocabSize)/2)))
# K ~ sqrt(n/2) this is a rule of thumb for choosing K,
# where n is the number of words in the model
# feel free to choose K with a fancier algorithm
dfW2V = wordVectorsDF.select('vector').withColumnRenamed('vector','features')
kmeans = KMeans(k=K, seed=1)
modelK = kmeans.fit(dfW2V)
labelsDF = modelK.transform(dfW2V).select('prediction').withColumnRenamed('prediction','labels')
vocabSize = wordVectorsDF.count()
timeKmeans = time.time() - t0
sc.stop()
## Print Some Results
printResults = 1 # set t
if (printResults):
## Read Tweets
print "="*80
print "Read Tweets..."
onehotenc = OneHotEncoder(inputCol=c, outputCol=c+"-onehot", dropLast=False)
newdf = onehotenc.transform(newdf).drop(c)
newdf = newdf.withColumnRenamed(c+"-onehot", c)
return newdf
dfhot = oneHotEncodeColumns(dfnumeric, ["Take-out","GoodFor_lunch", "GoodFor_dinner", "GoodFor_breakfast"])
dfhot.show(5)
# Taining set
assembler = VectorAssembler(inputCols = list(set(dfhot.columns) | set(['stars','review_count'])), outputCol="features")
train = assembler.transform(dfhot)
# Kmeans set for 5 clusters
knum = 5
kmeans = KMeans(featuresCol=assembler.getOutputCol(), predictionCol="cluster", k=knum, seed=0)
model = kmeans.fit(train)
print "Model Created!"
# See cluster centers:
centers = model.clusterCenters()
print("Cluster Centers: ")
for center in centers:
print(center)
# Apply the clustering model to our data:
prediction = model.transform(train)
prediction.groupBy("cluster").count().orderBy("cluster").show()
# Look at the features of each cluster
customerCluster = {}
# $example off$
from pyspark.sql import SparkSession
if __name__ == "__main__":
spark = SparkSession\
.builder\
.appName("KMeansExample")\
.getOrCreate()
# $example on$
# Loads data.
dataset = spark.read.format("libsvm").load("data/mllib/sample_kmeans_data.txt")
# Trains a k-means model.
kmeans = KMeans().setK(2).setSeed(1)
model = kmeans.fit(dataset)
# Make predictions
predictions = model.transform(dataset)
# Evaluate clustering by computing Silhouette score
evaluator = ClusteringEvaluator()
silhouette = evaluator.evaluate(predictions)
print("Silhouette with squared euclidean distance = " + str(silhouette))
# Shows the result.
centers = model.clusterCenters()
print("Cluster Centers: ")
for center in centers:
# Sukurtas `vecAssembler` objektas (kurio tipas yra `pyspark.ml.feature.VectorAssembler`) turi `transform` metodą, kuriuo iš stulpelių `x` ir `y` bus pagamintas vektorių stulpelis `features` ir iš jam perduodamo `DataFrame` objekto bus sukurtas naujas `DataFrame` su `features` stulpeliu:
# In[61]:
ca1mlFeaturizedDF = vecAssembler.transform(ca1MlDF)
ca1mlFeaturizedDF.show(5)
# Sukursime `pyspark.ml.clustering.KMeans` objektą (atminkime, kad 1 pavyzdyje naudojome `pyspark.mllib.clustering.KMeans`), kuriuo apmokysime `pyspark.ml.clustering.KMeansModel`. Duomenis pateiksime nebe `RDD`, o `DataFrame` objektu.
# In[62]:
from pyspark.ml.clustering import KMeans as MlKMeans
firstMlKMeans = MlKMeans(
featuresCol="features", predictionCol="prediction", k=2,
initMode="k-means||", maxIter=20)
type(firstMlKMeans)
# `pyspark.ml` paketo modelių klasės turi `explainParams` metodą, kuruo išvedami modelio parametrų paaiškinimai.
# In[63]:
print(firstMlKMeans.explainParams())
# Apmokykime modelį.
# In[64]:
def test_kmeans_param(self):
algo = KMeans()
self.assertEqual(algo.getInitMode(), "k-means||")
algo.setK(10)
self.assertEqual(algo.getK(), 10)
algo.setInitSteps(10)
self.assertEqual(algo.getInitSteps(), 10)
self.assertEqual(algo.getDistanceMeasure(), "euclidean")
algo.setDistanceMeasure("cosine")
self.assertEqual(algo.getDistanceMeasure(), "cosine")
print(colStdDev)
#Place the means and std.dev values in a broadcast variable
bcMeans = sc.broadcast(colMeans)
bcStdDev = sc.broadcast(colStdDev)
csAuto = autoVector.map(centerAndScale)
#csAuto.collect()
#csAuto.foreach(println)
print(csAuto)
#Create Spark Data Frame
autoRows = csAuto.map(lambda f:Row(features=f))
autoDf = SQLContext.createDataFrame(autoRows)
autoDf.select("features").show(10)
kmeans = KMeans(k=3, seed=1)
model = kmeans.fit(autoDf)
predictions = model.transform(autoDf)
predictions.collect()
predictions.foreach(println)
#Plot the results in a scatter plot
unstripped = predictions.map(unstripData)
predList=unstripped.collect()
predPd = pd.DataFrame(predList)
# preparing to save the clustered data
list_current_gni_final_maped = current_gni_final_maped.collect()
list_current_gni_rdd = current_gni_rdd.collect()
list_predictions_pandas=predictions.toPandas()
list_predictions_temp=list_predictions_pandas.as_matrix()