def evaluate(predictions):
"""
Evaluation Metrics
"""
# label to indexedLabel mappings
# out = sorted(set([(i[0], i[1]) for i in predictions.select(predictions.label, predictions.indexedLabel).collect()]), key=lambda x: x[0])
print "Predictions"
predictions.select("prediction", "indexedLabel", "features").show(5)
# Select (prediction, true label) and evaluate model
predictionAndLabels = predictions.select("prediction", "indexedLabel").rdd
metrics = MulticlassMetrics(predictionAndLabels)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
# Statistics by class
labels = predictions.map(lambda lp: lp.label).distinct().collect()
for label in sorted(labels):
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Measure = %s" % (label, metrics.fMeasure(label, beta=1.0)))
# Weighted stats
print("Weighted recall = %s" % metrics.weightedRecall)
print("Weighted precision = %s" % metrics.weightedPrecision)
print("Weighted F(1) Score = %s" % metrics.weightedFMeasure())
print("Weighted F(0.5) Score = %s" % metrics.weightedFMeasure(beta=0.5))
print("Weighted false positive rate = %s" % metrics.weightedFalsePositiveRate)
treeModel = model.stages[2]
print treeModel # summary only
def evaluate(labelsAndPredictions, data, labels):
"""
Evaluation Metrics
"""
# Instantiate metrics object
metrics = MulticlassMetrics(labelsAndPredictions)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
# Statistics by class
for label in sorted(labels):
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Measure = %s" % (label, metrics.fMeasure(label, beta=1.0)))
# Weighted stats
print("Weighted recall = %s" % metrics.weightedRecall)
print("Weighted precision = %s" % metrics.weightedPrecision)
print("Weighted F(1) Score = %s" % metrics.weightedFMeasure())
print("Weighted F(0.5) Score = %s" % metrics.weightedFMeasure(beta=0.5))
print("Weighted false positive rate = %s" % metrics.weightedFalsePositiveRate)
def NaiveBayesEvaluation(TransformedDataset):
nb = NaiveBayes()
nb.setLabelCol("LabelIndex")
nb.setPredictionCol("Label_Prediction")
training, test = TransformedDataset.randomSplit([0.8, 0.2], seed=11)
nvModel = nb.fit(training)
prediction = nvModel.transform(test)
# selected = prediction.select("body", "LabelIndex", "label", "Label_Prediction")
# for row in selected.collect():
# print(row)
from pyspark.mllib.evaluation import MulticlassMetrics
predictionAndLabels = prediction.select(
"Label_Prediction",
"LabelIndex").rdd.map(lambda r: (float(r[0]), float(r[1])))
# predictionAndLabels = test.rdd.map(lambda lp: (float(nvModel.predict(lp.features)), lp.label))
metrics = MulticlassMetrics(predictionAndLabels)
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
# Statistics by class
labels = prediction.rdd.map(lambda lp: lp.label).distinct().collect()
labelIndices = prediction.rdd.map(
lambda lp: lp.LabelIndex).distinct().collect()
labelIndicesPairs = prediction.rdd.map(
lambda lp: (lp.label, lp.LabelIndex)).distinct().collect()
print("Labels", labels)
print("Label Indices", labelIndices)
print("Label Indice Pairs", labelIndicesPairs)
for label, labelIndex in sorted(labelIndicesPairs):
print("\n Class %s precision = %s" %
(label, metrics.precision(labelIndex)))
print("Class %s recall = %s" % (label, metrics.recall(labelIndex)))
print(
"Class %s F1 Measure = %s" %
(label, metrics.fMeasure(labelIndex, beta=1.0)), "\n")
# Weighted stats
print("Weighted recall = %s" % metrics.weightedRecall)
print("Weighted precision = %s" % metrics.weightedPrecision)
print("Weighted F(1) Score = %s" % metrics.weightedFMeasure())
print("Weighted F(0.5) Score = %s" % metrics.weightedFMeasure(beta=0.5))
print("Weighted false positive rate = %s" %
metrics.weightedFalsePositiveRate)
def evaluate_predictions(predictions, show=True):
from pyspark.ml.evaluation import BinaryClassificationEvaluator
from pyspark.mllib.evaluation import BinaryClassificationMetrics, MulticlassMetrics
log = {}
evaluator = BinaryClassificationEvaluator(metricName='areaUnderROC')
log['auroc'] = evaluator.evaluate(predictions)
# Show Validation Score (AUPR)
evaluator = BinaryClassificationEvaluator(metricName='areaUnderPR')
log['aupr'] = evaluator.evaluate(predictions)
# Metrics
predictionRDD = predictions.select(
['label', 'prediction']).rdd.map(lambda line: (line[1], line[0]))
metrics = MulticlassMetrics(predictionRDD)
# Overall statistics
log['precision'] = metrics.precision()
log['recall'] = metrics.recall()
log['F1 Measure'] = metrics.fMeasure()
# Statistics by class
distinctPredictions = collect_tuple(
predictions.select('prediction').distinct())
for x in sorted(distinctPredictions):
log[x] = {}
log[x]['precision'] = metrics.precision(x)
log[x]['recall'] = metrics.recall(x)
log[x]['F1 Measure'] = metrics.fMeasure(x, beta=1.0)
# Confusion Matrix
log['cm'] = metrics.confusionMatrix().toArray()
log['cmpercent'] = cm_percent(log['cm'], predictions.count(), show)
if show:
show_predictions(predictions)
print('Confusion Matrix')
print(' TP', 'FN\n', 'FP', 'TN')
print(log['cm'])
print(' PC', 'FN\n', 'FP', 'PW')
print(log['cmpercent'])
print('')
print("Area under ROC = {}".format(log['auroc']))
print("Area under AUPR = {}".format(log['aupr']))
print('\nOverall\ntprecision = {}\nrecall = {}\nF1 Measure = {}\n'.
format(log['precision'], log['recall'], log['F1 Measure']))
for x in sorted(distinctPredictions):
print('Label {}\ntprecision = {}\nrecall = {}\nF1 Measure = {}\n'.
format(x, log[x]['precision'], log[x]['recall'],
log[x]['F1 Measure']))
return log
def printMetrics(predictions_and_labels, output_file):
metrics = MulticlassMetrics(predictions_and_labels)
output_file.write('Precision of True '+str(metrics.precision(1))+'\n')
output_file.write('Precision of False' + str(metrics.precision(0))+'\n')
output_file.write('Recall of True '+str(metrics.recall(1))+'\n')
output_file.write('Recall of False '+str(metrics.recall(0))+'\n')
output_file.write('F-1 Score '+str(metrics.fMeasure())+'\n')
output_file.write('Confusion Matrix\n'+str(metrics.confusionMatrix().toArray())+'\n')
print('Precision of True '+str(metrics.precision(1)))
print('Precision of False'+str(metrics.precision(0)))
print('Recall of True '+str(metrics.recall(1)))
print('Recall of False '+str(metrics.recall(0)))
print('F-1 Score '+str(metrics.fMeasure()))
print('Confusion Matrix\n'+str(metrics.confusionMatrix().toArray()))
def main(spark, model_file, data_file):
'''Main routine for supervised evaluation
Parameters
----------
spark : SparkSession object
model_file : string, path to store the serialized model file
data_file : string, path to the parquet file to load
'''
###
# TODO: YOUR CODE GOES HERE
#load best lr model
model = PipelineModel.load(model_file)
# Load the test dataframe
test = spark.read.parquet(data_file)
predictions = model.transform(test)
predictionAndLabels = predictions.rdd.map(lambda lp:
(lp.prediction, lp.label))
metrics = MulticlassMetrics(predictionAndLabels)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
print("Overall Stats:")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
# Weighted stats
print("Weighted precision = %s" % metrics.weightedPrecision)
print("Weighted recall = %s" % metrics.weightedRecall)
print("Weighted F1 Score = %s" % metrics.weightedFMeasure())
# Statistics by class
print("Stats by class")
for (genre, label) in predictions.select('genre',
'label').distinct().collect():
print("Class %s precision = %s" % (genre, metrics.precision(label)))
print("Class %s recall = %s" % (genre, metrics.recall(label)))
print("Class %s F1 Score = %s" %
(genre, metrics.fMeasure(label, beta=1.0)))
def multi_clf_performance(name, method, train, test):
model = method.fit(train)
prediction = model.transform(test)
print(f"-----------Performance of {name} on testing set-----------")
# Compute raw scores on the test set
predictionAndLabels = prediction.select('prediction', 'label')
# Instantiate metrics object
metrics = MulticlassMetrics(predictionAndLabels.rdd)
# Overall statistics
print("----------Summary Stats----------------------")
print(f"Weighted precision: {multi_evaluator.evaluate(prediction, {multi_evaluator.metricName: 'weightedPrecision'})}")
print(f"Weighted recall: {multi_evaluator.evaluate(prediction, {multi_evaluator.metricName: 'weightedRecall'})}")
print(f"F1 Score: {multi_evaluator.evaluate(prediction, {multi_evaluator.metricName: 'f1'})}")
print(f"Accuracy: {multi_evaluator.evaluate(prediction, {multi_evaluator.metricName: 'accuracy'})}")
# Statistics by class
print("--------Stats by class----------------------")
labels = [row.asDict()['label'] for row in test.select('label').distinct().collect()]
for label in sorted(labels):
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Score = %s" % (label, metrics.fMeasure(label, beta=1.0)))
# Weighted stats
#print("--------Weighted Stats----------------------")
#print("Weighted precision = %s" % metrics.weightedPrecision)
#print("Weighted recall = %s" % metrics.weightedRecall)
#print("Weighted F1 Score = %s" % metrics.weightedFMeasure())
#print("Weighted false positive rate = %s" % metrics.weightedFalsePositiveRate)
print("-----------------------------------------------------------")
def calculate_metrics(self, df):
"""
define your own metrics to evaluate cross validation
:params:
df: dataframe containing {aprediction} and {label} columns
:returns:
confusion matrix
"""
# turn gt into label
preds_and_labels = df.select('prediction',
f.col('label').cast(t.FloatType()))
metrics = MulticlassMetrics(preds_and_labels.rdd.map(tuple))
# confusion matrix
metrics_dict = dict(
# unweighted measures
tpr=metrics.truePositiveRate(label=1.0),
fpr=metrics.falsePositiveRate(label=1.0),
precision=metrics.precision(label=1.0),
recall=metrics.recall(label=1.0),
fMeasure=metrics.fMeasure(label=1.0))
metrics_dict = {
k: round(v, 3) if k != "confusion" else v
for k, v in metrics_dict.items()
}
return metrics_dict
def evaluate(df, labelCols, gettopX=-1, getfirstX=-1):
labelCols2 = [i + "_pred" for i in labelCols]
df.cache()
r_list = {
i: np.zeros((len(labelCols)))
for i in ['accuracy', 'precision', 'recall', 'fmeasure']
}
for i in xrange(len(labelCols)):
predandlabels = df.select(labelCols2[i], labelCols[i]).rdd \
.map(lambda x: (float(x[labelCols2[i]]), float(x[labelCols[i]])))
metrics = MulticlassMetrics(predandlabels)
# print metrics.confusionMatrix()
r_list['accuracy'][i] = metrics.accuracy
r_list['precision'][i] = metrics.precision(1.0)
r_list['recall'][i] = metrics.recall(1.0)
r_list['fmeasure'][i] = metrics.fMeasure(label=1.0)
results = {}
for m, rs in r_list.iteritems():
results[m] = np.mean(rs)
for code, num in [('top', gettopX), ('first', getfirstX)]:
if num <= 0: continue
if code == 'top':
idx = np.argsort(np.nan_to_num(r_list['fmeasure']))[-num:]
elif code == 'first':
idx = xrange(num)
for m, rs in r_list.iteritems():
results['{0}_{1}'.format(m, code)] = np.mean(rs[idx])
return results
def print_performance_metrics(predictions):
# Evaluate model
evaluator = BinaryClassificationEvaluator(rawPredictionCol="rawPrediction")
auc = evaluator.evaluate(predictions,
{evaluator.metricName: "areaUnderROC"})
aupr = evaluator.evaluate(predictions,
{evaluator.metricName: "areaUnderPR"})
print("auc = {}".format(auc))
print("aupr = {}".format(aupr))
# Get RDD of predictions and labels for eval metrics
predictionAndLabels = predictions.select("prediction", "label").rdd
# Instantiate metrics objects
binary_metrics = BinaryClassificationMetrics(predictionAndLabels)
multi_metrics = MulticlassMetrics(predictionAndLabels)
# Area under precision-recall curve
print("Area under PR = {}".format(binary_metrics.areaUnderPR))
# Area under ROC curve
print("Area under ROC = {}".format(binary_metrics.areaUnderROC))
# Accuracy
print("Accuracy = {}".format(multi_metrics.accuracy))
# Confusion Matrix
print(multi_metrics.confusionMatrix())
# F1
print("F1 = {}".format(multi_metrics.fMeasure(1.0)))
# Precision
print("Precision = {}".format(multi_metrics.precision(1.0)))
# Recall
print("Recall = {}".format(multi_metrics.recall(1.0)))
# FPR
print("FPR = {}".format(multi_metrics.falsePositiveRate(1.0)))
# TPR
print("TPR = {}".format(multi_metrics.truePositiveRate(1.0)))
def getF1Score(model, test_df):
pred = model.transform(test_df)
pl = pred.select("label", "prediction").rdd.cache()
metrics = MulticlassMetrics(pl)
f1score = metrics.fMeasure()
print("the F1-score of the model is : {}".format(f1score))
return f1score
def printMeasurementMetrics(predictions_and_labels):
metrics = MulticlassMetrics(predictions_and_labels)
print('Precision Result of setosa: ', metrics.precision(1))
print('Precision Result of versicolor:', metrics.precision(2))
print('Precision Result of virginica:', metrics.precision(3))
print('F-1 Score: ', metrics.fMeasure())
print('Confusion Matrix\n', metrics.confusionMatrix().toArray())
def evaluate(model, word_column="words", vectorizer="w2v"):
doc2vecs_df = featurize(word_column, vectorizer)
if type(model) == LinearSVC:
paramGrid = ParamGridBuilder() \
.addGrid(model.regParam, [0.1]) \
.build()
elif type(model) == GBTClassifier:
paramGrid = ParamGridBuilder() \
.addGrid(model.maxIter, [50]) \
.build()
elif type(model) == RandomForestClassifier:
paramGrid = ParamGridBuilder() \
.addGrid(model.maxBins, [100]) \
.build()
elif type(model) == MultilayerPerceptronClassifier:
paramGrid = ParamGridBuilder() \
.addGrid(model.layers, [[122, 50, 2]]) \
.build()
# .addGrid(model.layers, [[120, 2], [120, 50, 2], [120, 75, 50, 2]]) \
elif type(model) == FMClassifier:
paramGrid = ParamGridBuilder() \
.addGrid(model.stepSize, [.01, .001]) \
.build()
print('Evaluating...')
w2v_train_df, w2v_test_df = doc2vecs_df.randomSplit([0.8, 0.2])
si = StringIndexer(inputCol="LABEL", outputCol="label")
model_evaluator = MulticlassClassificationEvaluator(
labelCol="label", predictionCol="prediction", metricName="f1")
classifier_pipeline = Pipeline(stages=[si, model])
crossval = CrossValidator(estimator=classifier_pipeline,
estimatorParamMaps=paramGrid,
evaluator=model_evaluator,
numFolds=5)
fit_model = crossval.fit(doc2vecs_df)
predictions = fit_model.transform(w2v_test_df)
# predictions.toPandas().to_csv('predictions.csv')
# predictions.groupBy('prediction', 'label', 'PRODUCT_CATEGORY')
# predictions.describe()
summarizer = Summarizer.metrics("mean", "count")
predictions.select(
summarizer.summary(predictions.filter(
predictions.label == 1).pos)).show(truncate=False)
preds_and_labels = predictions.select(['prediction', 'label'])
metrics = MulticlassMetrics(preds_and_labels.rdd.map(tuple))
print('Confusion Matrix')
print(metrics.confusionMatrix().toArray())
# Overall statistics
precision = metrics.precision(1.0)
recall = metrics.recall(1.0)
f1Score = metrics.fMeasure(1.0)
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
accuracy = model_evaluator.evaluate(predictions)
trainingSummary = fit_model.bestModel.stages[-1].extractParamMap()
print(trainingSummary)
return accuracy
def printMetrics(predictions_and_labels):
metrics = MulticlassMetrics(predictions_and_labels)
print 'Precision of True ', metrics.precision(1)
print 'Precision of False', metrics.precision(0)
print 'Recall of True ', metrics.recall(1)
print 'Recall of False ', metrics.recall(0)
print 'F-1 Score ', metrics.fMeasure()
print 'Confusion Matrix\n', metrics.confusionMatrix().toArray()
def printMetrics(result):
metrics = MulticlassMetrics(result)
print("\nPrecision of True\n", metrics.precision(1))
print("\nPrecision of False\n", metrics.precision(0))
print("\nRecall of True\n", metrics.recall(1))
print("\nRecall of False\n", metrics.recall(0))
print("\nF1 score\n", metrics.fMeasure())
print("\nConfusion Matrix\n", metrics.confusionMatrix().toArray())
def evaluate(df_prediction):
evaluator = BinaryClassificationEvaluator()
rc = evaluator.evaluate(df_prediction, {evaluator.metricName: "areaUnderROC"})
pr = evaluator.evaluate(df_prediction, {evaluator.metricName: "areaUnderPR"})
predictionRDD = df_prediction.select(['label', 'prediction']).rdd.map(lambda line: (line[1], line[0]))
metrics = MulticlassMetrics(predictionRDD)
f1 = metrics.fMeasure()
return [roc,pr,f1]
def displayMetrics(pred):
ev = MulticlassMetrics(pred.select(["label", "prediction"]).rdd)
# Overall statistics
print("Accuracy = %s" % ev.accuracy)
print("Precision = %s" % ev.precision())
print("Recall = %s" % ev.recall())
print("F1 Score = %s" % ev.fMeasure())
def main(spark, model_file, data_file):
'''Main routine for supervised evaluation
Parameters
----------
spark : SparkSession object
model_file : string, path to store the serialized model file
data_file : string, path to the parquet file to load
'''
# Load data.
dataset = spark.read.parquet(data_file)
# Load model.
model = PipelineModel.load(model_file)
prediction = model.transform(dataset)
predictionAndLabels = prediction.select(["prediction", "label"]).rdd
# Instantiate metrics object
metrics = MulticlassMetrics(predictionAndLabels)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
print("\n")
# Weighted stats
print("Weighted recall = %s" % metrics.weightedRecall)
print("Weighted precision = %s" % metrics.weightedPrecision)
print("Weighted F(1) Score = %s" % metrics.weightedFMeasure())
print("\n")
labels = predictionAndLabels.map(lambda lp: lp.label).distinct().collect()
for label in sorted(labels):
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Measure = %s" %
(label, metrics.fMeasure(label, beta=1.0)))
def evaluate(pl): # input predictionsAndLabels
testErr = predictionsAndLabels.filter(
lambda lp: lp[0] != lp[1]).count() / float(predictionsAndLabels.count())
metrics = MulticlassMetrics(predictionsAndLabels)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
return testErr, precision, recall, f1Score
def metrics_basic(data):
metrics = MulticlassMetrics(data)
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
def printMetrics(self, preds, prediction="indexedLabel", indexedLabel="prediction"):
metrics = MulticlassMetrics(preds.select(prediction, indexedLabel).rdd)
labels = [0, 1]
for label in sorted(labels):
try:
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Measure = %s" % (label, metrics.fMeasure(label, beta=1.0)))
except:
print("No malicious predictions")
def classification_report(actual_data, prediction_data):
# Calculate actual / predicted labels in rdd from
prediction_and_labels = prepare_data(actual_data, prediction_data)
# Calculate calculate class level metrics
metrics = MulticlassMetrics(prediction_and_labels)
classes = set(actual_data.rdd.map(lambda x: x.test_label[0]).collect())
print('Class\tPrecision\tRecall\tF-Score')
for c in sorted(classes):
print('{}\t{}\t{}\t{}'.format(c, round(metrics.precision(c), 3),
round(metrics.recall(c), 3),
round(metrics.fMeasure(c), 3)))
def evaluate(predictionAndLabels):
log = {}
# Show Validation Score (AUROC)
evaluator = BinaryClassificationEvaluator(metricName='areaUnderROC')
log['AUROC'] = "%f" % evaluator.evaluate(predictionAndLabels)
print("Area under ROC = {}".format(log['AUROC']))
# Show Validation Score (AUPR)
evaluator = BinaryClassificationEvaluator(metricName='areaUnderPR')
log['AUPR'] = "%f" % evaluator.evaluate(predictionAndLabels)
print("Area under PR = {}".format(log['AUPR']))
# Metrics
predictionRDD = predictionAndLabels.select(['label', 'prediction']) \
.rdd.map(lambda line: (line[1], line[0]))
metrics = MulticlassMetrics(predictionRDD)
# Confusion Matrix
print(metrics.confusionMatrix().toArray())
# Overall statistics
log['precision'] = "%s" % metrics.precision()
log['recall'] = "%s" % metrics.recall()
log['F1 Measure'] = "%s" % metrics.fMeasure()
print("[Overall]\tprecision = %s | recall = %s | F1 Measure = %s" % \
(log['precision'], log['recall'], log['F1 Measure']))
# Statistics by class
labels = [0.0, 1.0]
for label in sorted(labels):
log[label] = {}
log[label]['precision'] = "%s" % metrics.precision(label)
log[label]['recall'] = "%s" % metrics.recall(label)
log[label]['F1 Measure'] = "%s" % metrics.fMeasure(label, beta=0.5)
print("[Class %s]\tprecision = %s | recall = %s | F1 Measure = %s" \
% (label, log[label]['precision'],
log[label]['recall'], log[label]['F1 Measure']))
return log
def evaluateClassification(self, predictionAndLabels):
metrics = MulticlassMetrics(predictionAndLabels)
cm = metrics.confusionMatrix()
result = {}
result['Matrix'] = cm.toArray().tolist()
result['Precision'] = metrics.precision()
result['Recall'] = metrics.recall()
result['F1 Score'] = metrics.fMeasure()
return result
def overall_report(actual_data, prediction_data):
# Calculate actual / predicted labels in rdd from
prediction_and_labels = prepare_data(actual_data, prediction_data)
# Calculate actual / predicted labels in rdd from
metrics = MulticlassMetrics(prediction_and_labels)
# Calculate overall level metrics
# print('Precision:', metrics.precision(), type(metrics.precision()))
return sc.parallelize([
(Vectors.dense(metrics.precision()), Vectors.dense(metrics.recall()),
Vectors.dense(metrics.fMeasure()))
]).toDF(['Precision', 'Recall', 'F-Score'])
def main():
sc = SparkContext(appName="BayesClassifer")
htf = HashingTF(50000)
data = sc.textFile('/home/varshav/work/PycharmProjects/Sentiment/cleaned_bayes_labels.csv')
data_cleaned = data.map(lambda line : line.split(","))
# Create an RDD of LabeledPoints using category labels as labels and tokenized, hashed text as feature vectors
data_hashed = data_cleaned.map(lambda (label, text): LabeledPoint(label, htf.transform(text)))
data_hashed.persist()
# data = sc.textFile('/home/admin/work/spark-1.4.1-bin-hadoop2.4/data/mllib/sample_naive_bayes_data.txt').map(parseLine)
#print data
# Split data aproximately into training (60%) and test (40%)
training, test = data_hashed.randomSplit([0.70, 0.30], seed=0)
sameModel = NaiveBayesModel.load(sc, "/home/varshav/work/PycharmProjects/StockAnalysis/myModel")
print "----------"
print sameModel.predict(htf.transform("posts jump in net profit"))
predictionAndLabel = test.map(lambda p: (sameModel.predict(p.features), p.label))
predictionAndLabel1 = training.map(lambda p: (sameModel.predict(p.features), p.label))
prediction = 1.0 * predictionAndLabel.filter(lambda (x, v): x == v).count() / test.count()
prediction1 = 1.0 * predictionAndLabel1.filter(lambda (x, v): x == v).count() / training.count()
buy_buy = 1.0 * predictionAndLabel.filter(lambda (x, v): x == 1 and v ==1).count()
# Instantiate metrics object
# Instantiate metrics object
metrics = MulticlassMetrics(predictionAndLabel)
# Overall statistics
precision = metrics.precision()
precision = normalize(precision)
recall = metrics.recall()
recall = normalize(recall)
f1Score = metrics.fMeasure()
f1Score = normalize(f1Score)
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
'''
# Statistics by class
labels = data_hashed.map(lambda lp: lp.label).distinct().collect()
for label in sorted(labels):
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Measure = %s" % (label, metrics.fMeasure(label, beta=1.0)))
'''
'''
def classification_report(actual_data, prediction_data):
# Calculate actual / predicted labels in rdd from
prediction_and_labels = prepare_data(actual_data, prediction_data)
# Calculate calculate class level metrics
metrics = MulticlassMetrics(prediction_and_labels)
classes = set(actual_data.rdd.map(lambda x: x.test_labels[0]).collect())
results = [(Vectors.dense(float(c)),
Vectors.dense(round(metrics.precision(c), 3)),
Vectors.dense(round(metrics.recall(c), 3)),
Vectors.dense(round(metrics.fMeasure(c), 3)))
for c in sorted(classes)]
return sc.parallelize(results).toDF(
['Class', 'Precision', 'Recall', 'F-Score'])
def validate_tffm(spark, sc, model, test_df, s3_metrics_path, s3_endpoint_path):
# get predictions
validation_df = model.transform(test_df)
metricsSchema = StructType() \
.add("metric", StringType()) \
.add("value", DoubleType())
metrics_names = []
# apply threshold
def thresholdScore(x):
retval = 0.0
if x > 0.5:
retval = 1.0
return retval
thresholdScoreUdf = F.UserDefinedFunction(thresholdScore, T.FloatType())
validation_df_round = validation_df.withColumn('rscore', thresholdScoreUdf(validation_df.score))
predTffm = validation_df_round.select(['label','rscore'])
predictionAndLabelsTffm = predTffm.rdd.map(lambda lp: (lp.rscore, lp.label))
metricsTffm = BinaryClassificationMetrics(predictionAndLabelsTffm)
metrics_names.append(("Area_under_PR",metricsTffm.areaUnderPR))
metrics_names.append(("Area_under_ROC",metricsTffm.areaUnderROC))
mmetricsTffm = MulticlassMetrics(predictionAndLabelsTffm)
metrics_names.append(("Precision",mmetricsTffm.precision()))
metrics_names.append(("Recall",mmetricsTffm.recall()))
metrics_names.append(("F1",mmetricsTffm.fMeasure()))
metrics_names.append(("Weighted_recall",mmetricsTffm.weightedRecall))
metrics_names.append(("Weighted_precision",mmetricsTffm.weightedPrecision))
metrics_names.append(("Weighted_F1",mmetricsTffm.weightedFMeasure()))
metrics_names.append(("Weighted_F05",mmetricsTffm.weightedFMeasure(beta=0.5)))
metrics_names.append(("Weighted_FP_rate",mmetricsTffm.weightedFalsePositiveRate))
mRdd = sc.parallelize(metrics_names).coalesce(1)
dfMetrics = spark.createDataFrame(mRdd, metricsSchema)
dfMetrics.write.csv("{0}/{1}".format(s3_metrics_path, model.endpointName), mode="overwrite")
endpointSchema = StructType() \
.add("time", StringType()) \
.add("endpoint", StringType())
endpoint_name = []
endpoint_name.append((str(time.time()),str(model.endpointName)))
eRdd = sc.parallelize(endpoint_name).coalesce(1)
dfEndpoint = spark.createDataFrame(eRdd, endpointSchema)
dfEndpoint.write.csv("{0}/endpoint.txt".format(s3_endpoint_path), mode="overwrite")
def generateJson(AlgorithmName, taskid, traindata, predictionAndLabels):
jsonContent = dict()
jsonContent['AlgorithmName'] = AlgorithmName
jsonContent['TaskId'] = taskid
labels = traindata.map(lambda lp: lp.label).distinct().collect()
jsonContent['LabelNum'] = len(labels)
metrics = MulticlassMetrics(predictionAndLabels)
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
confusion_matrix = metrics.confusionMatrix().toArray()
jsonContent['Precision'] = precision
jsonContent['Recall'] = recall
jsonContent['F1Score'] = f1Score
jsonContent['ConfusionMatrix'] = confusion_matrix.tolist()
jsonContent['Labels'] = list()
for label in sorted(labels):
tempList = dict()
tempList['Precision'] = metrics.precision(label)
tempList['Recall'] = metrics.recall(label)
tempList['F1Measure'] = metrics.fMeasure(label, beta=1.0)
jsonContent['Labels'].append(tempList)
jsonContent['WeightedStats'] = dict()
jsonContent['WeightedStats']['Precision'] = metrics.weightedRecall
jsonContent['WeightedStats']['F1Score'] = metrics.weightedFMeasure()
jsonContent['WeightedStats']['FalsePositiveRate'] = metrics.weightedFalsePositiveRate
with open(taskid + '.json', 'w') as jsonFile:
json.dump(jsonContent, jsonFile, indent=4, separators=(',', ': '))
jsonFile.flush()
def performance(predictions):
predictionRDD = predictions.select(['label', 'prediction']).rdd.map(lambda line: (line[1], line[0]))
binmetrics = BinaryClassificationMetrics(predictionRDD)
metrics = MulticlassMetrics(predictionRDD)
results = {'predictions':predictions,
'areaUnderROC':binmetrics.areaUnderROC,
'areaUnderPR':binmetrics.areaUnderPR,
'confusionMatrix':metrics.confusionMatrix().toArray(),
'accuracy':metrics.accuracy,
'precision':metrics.precision(),
'recall':metrics.recall(),
'f1measure':metrics.fMeasure()}
return results
def overall_report(actual_data, prediction_data):
# Calculate actual / predicted labels in rdd from
prediction_and_labels = prepare_data(actual_data, prediction_data)
# Calculate actual / predicted labels in rdd from
metrics = MulticlassMetrics(prediction_and_labels)
# Calculate overall level metrics
# print('Precision:', metrics.precision(), type(metrics.precision()))
# return sc.parallelize([(Vectors.dense(metrics.accuracy),
# Vectors.dense(metrics.precision()),
# Vectors.dense(metrics.recall()),
# Vectors.dense(metrics.fMeasure()))]).toDF(['Accuracy', 'Precision', 'Recall', 'F - Score'])
print('Accuracy\tPrecision\tRecall\tF-Score')
print('{}\t{}\t{}\t{}'.format(metrics.accuracy, metrics.precision(),
metrics.recall(), metrics.fMeasure()))
def performancerdd(self):
self.calculator = 'RDDs'
print('Calculating performance metrics using RDDs...')
predictionRDD = self.predictions.select(['label','prediction']).rdd.map(lambda line: (line[1],line[0]))
binmetrics = BinaryClassificationMetrics(predictionRDD)
metrics = MulticlassMetrics(predictionRDD)
self.areaUnderROC = binmetrics.areaUnderROC
self.areaUnderPR = binmetrics.areaUnderPR
self.confusionMatrix = metrics.confusionMatrix().toArray()
self.accuracy = metrics.accuracy
self.precision = metrics.precision()
self.recall = metrics.recall()
self.f1measure = metrics.fMeasure()
self.falsePositive = metrics.falsePositiveRate(1.0)
self.falseNegative = metrics.falsePositiveRate(0.0)
def printFinalResultMetrics(predictions_and_labels):
metrics = MulticlassMetrics(predictions_and_labels)
print '\n'
print 'Precision of Setosa ', metrics.precision(1)
print 'Precision of Versicolor', metrics.precision(2)
print 'Precision of Virginica', metrics.precision(3)
print '\n'
print 'Recall of Setosa ', metrics.recall(1)
print 'Recall of Versicolor ', metrics.recall(2)
print 'Recall of Virginica ', metrics.recall(3)
print '\n'
print 'F-1 Score ', metrics.fMeasure()
print '\n\n'
print 'Confusion Matrix\n', metrics.confusionMatrix().toArray()
print '\n\n'
return
def modelStatistics(labelsAndPredictions):
metrics = MulticlassMetrics(labelsAndPredictions)
print(metrics.confusionMatrix())
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
# Weighted stats
print("Weighted recall = %s" % metrics.weightedRecall)
print("Weighted precision = %s" % metrics.weightedPrecision)
print("Weighted F(1) Score = %s" % metrics.weightedFMeasure())
print("Weighted F(0.5) Score = %s" % metrics.weightedFMeasure(beta=0.5))
print("Weighted false positive rate = %s" % metrics.weightedFalsePositiveRate)
model = SVMWithSGD.train(trainParsed, iterations=100) # Training Error trainLabelsAndPreds = trainParsed.map(lambda p: (p.label, float(model.predict(p.features)))) trainErr = trainLabelsAndPreds.filter(lambda (v, p): v != p).count()/float(trainParsed.count()) print trainErr # Test Error testLabelsAndPreds = testParsed.map(lambda p: (p.label, float(model.predict(p.features)))) testErr = testLabelsAndPreds.filter(lambda (v, p): v != p).count()/float(testParsed.count()) print testErr metrics = BinaryClassificationMetrics(testLabelsAndPreds) print metrics.areaUnderROC print metrics.areaUnderPR mcMetrics = MulticlassMetrics(testLabelsAndPreds) #TODO: Do this for classes 1.0,0.0 and not just overall print mcMetrics.precision() print mcMetrics.recall() print mcMetrics.fMeasure() model.save(sc, "SVMModel") ### Run Model on Validation Set ## TODO: output file of zipcodes and predicted success metrics ## TODO: Use bokeh on file to make visualization of the US
def train_model (conf):
sc = SparkUtil.get_spark_context (conf.spark_conf)
conf.output_dir = conf.output_dir.replace ("file:", "")
conf.output_dir = "file://{0}".format (conf.output_dir)
labeled = Evaluate.load_all (sc, conf). \
map (lambda b : LabeledPoint ( label = 1.0 if b.fact else 0.0,
features = [ b.paraDist, b.sentDist, b.docDist ] ) )
# labeled = sc.parallelize ([ round ((x/10) * 9) for x in random.sample(range(1, 100000000), 30000) ]). \
# map (lambda b : LabeledPoint ( 1.0 if b % 2 == 0 else 0.0,
# [ b, b * 2, b * 9 ] ) )
# print (labeled.collect ())
train, test = labeled.randomSplit (weights=[ 0.8, 0.2 ], seed=12345)
count = train.count ()
start = time.time ()
model = LogisticRegressionWithLBFGS.train (train)
elapsed = time.time () - start
print ("Trained model on training set of size {0} in {1} seconds".format (count, elapsed))
start = time.time ()
model_path = os.path.join (conf.output_dir, "eval", "model")
file_path = model_path.replace ("file://", "")
if os.path.isdir (file_path):
print ("Removing existing model {0}".format (file_path))
shutil.rmtree (file_path)
model.save(sc, model_path)
sameModel = LogisticRegressionModel.load(sc, model_path)
elapsed = time.time () - start
print ("Saved and restored model to {0} in {1} seconds".format (model_path, elapsed))
# Metrics
labelsAndPreds = test.map (lambda p: (p.label, model.predict (p.features)))
trainErr = labelsAndPreds.filter(lambda (v, p): v != p).count () / float (train.count())
print("Training Error => {0}".format (trainErr))
predictionsAndLabels = labelsAndPreds.map (lambda x : ( float(x[1]), float(x[0]) ))
metrics = MulticlassMetrics (predictionsAndLabels)
print (" --------------> {0}".format (predictionsAndLabels.take (1000)))
#print (labelsAndPreds.collect ())
print ("\nMETRICS:")
try:
print ("false positive (0.0): {0}".format (metrics.falsePositiveRate(0.0)))
print ("false positive (1.0): {0}".format (metrics.falsePositiveRate(1.0)))
except:
traceback.print_exc ()
try:
print ("precision : {0}".format (metrics.precision(1.0)))
except:
traceback.print_exc ()
try:
print ("recall : {0}".format (metrics.recall(1.0)))
except:
traceback.print_exc ()
try:
print ("fMeasure : {0}".format (metrics.fMeasure(0.0, 2.0)))
except:
traceback.print_exc ()
print ("confusion matrix : {0}".format (metrics.confusionMatrix().toArray ()))
print ("precision : {0}".format (metrics.precision()))
print ("recall : {0}".format (metrics.recall()))
print ("weighted false pos : {0}".format (metrics.weightedFalsePositiveRate))
print ("weighted precision : {0}".format (metrics.weightedPrecision))
print ("weighted recall : {0}".format (metrics.weightedRecall))
print ("weight f measure : {0}".format (metrics.weightedFMeasure()))
print ("weight f measure 2 : {0}".format (metrics.weightedFMeasure(2.0)))
print ("")
# Regression metrics
predictedAndObserved = test.map (lambda p: (model.predict (p.features) / 1.0 , p.label / 1.0 ) )
regression_metrics = RegressionMetrics (predictedAndObserved)
print ("explained variance......: {0}".format (regression_metrics.explainedVariance))
print ("absolute error..........: {0}".format (regression_metrics.meanAbsoluteError))
print ("mean squared error......: {0}".format (regression_metrics.meanSquaredError))
print ("root mean squared error.: {0}".format (regression_metrics.rootMeanSquaredError))
print ("r2......................: {0}".format (regression_metrics.r2))
print ("")
labelsAndPreds = test.map (lambda p: (p.label, sameModel.predict (p.features)))
testErr = labelsAndPreds.filter (lambda (v, p): v != p).count () / float (test.count ())
print ("Testing Error => {0}".format (testErr))
training, test = data.randomSplit([0.6, 0.4], seed=11)
training.cache()
# Run training algorithm to build the model
model = LogisticRegressionWithLBFGS.train(training, numClasses=3)
# Compute raw scores on the test set
predictionAndLabels = test.map(lambda lp: (float(model.predict(lp.features)), lp.label))
# Instantiate metrics object
metrics = MulticlassMetrics(predictionAndLabels)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
# Statistics by class
labels = data.map(lambda lp: lp.label).distinct().collect()
for label in sorted(labels):
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Measure = %s" % (label, metrics.fMeasure(label, beta=1.0)))
# Weighted stats
print("Weighted recall = %s" % metrics.weightedRecall)
print("Weighted precision = %s" % metrics.weightedPrecision)
def logisticRegression(trainFile, testFile, taskid, sc):
# Load training data in LIBSVM format
trainData = MLUtils.loadLibSVMFile(sc, trainFile)
testData = MLUtils.loadLibSVMFile(sc, testFile)
# Split data into training (60%) and test (40%)
# traindata, testdata = data.randomSplit([0.6, 0.4], seed = 11L)
# traindata.cache()
# Load testing data in LIBSVM format
#testdata = MLUtils.loadLibSVMFile(sc, loadTestingFilePath)
labelNum = trainData.map(lambda lp: lp.label).distinct().count()
# Run training algorithm to build the model
model = LogisticRegressionWithLBFGS.train(trainData, numClasses=labelNum)
# Compute raw scores on the test set
predictionAndLabels = testData.map(lambda lp: (float(model.predict(lp.features)), lp.label))
Json.generateJson("LogisticRegression", taskid, trainData, predictionAndLabels);
# Instantiate metrics object
metrics = MulticlassMetrics(predictionAndLabels)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
#confusion_matrix = metrics.confusionMatrix().toArray()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
# Statistics by class
labels = trainData.map(lambda lp: lp.label).distinct().collect()
for label in sorted(labels):
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Measure = %s" % (label, metrics.fMeasure(label, beta=1.0)))
# Weighted stats
print("Weighted recall = %s" % metrics.weightedRecall)
print("Weighted precision = %s" % metrics.weightedPrecision)
print("Weighted F(1) Score = %s" % metrics.weightedFMeasure())
print("Weighted F(0.5) Score = %s" % metrics.weightedFMeasure(beta=0.5))
print("Weighted false positive rate = %s" % metrics.weightedFalsePositiveRate)
# #return model parameters
# res = [('1','Yes','TP Rate', metrics.truePositiveRate(0.0)),
# ('2','Yes','FP Rate', metrics.falsePositiveRate(0.0)),
# ('3','Yes','Precision', metrics.precision(0.0)),
# ('4','Yes','Recall', metrics.recall(0.0)),
# ('5','Yes','F-Measure', metrics.fMeasure(0.0, beta=1.0)),
# ('1','Yes','TP Rate', metrics.truePositiveRate(1.0)),
# ('2','Yes','FP Rate', metrics.falsePositiveRate(1.0)),
# ('3','Yes','Precision', metrics.precision(1.0)),
# ('4','Yes','Recall', metrics.recall(1.0)),
# ('5','Yes','F-Measure', metrics.fMeasure(1.0, beta=1.0)),
# ('1','Yes','TP Rate', metrics.truePositiveRate(2.0)),
# ('2','Yes','FP Rate', metrics.falsePositiveRate(2.0)),
# ('3','Yes','Precision', metrics.precision(2.0)),
# ('4','Yes','Recall', metrics.recall(2.0)),
# ('5','Yes','F-Measure', metrics.fMeasure(2.0, beta=1.0))]
# #save output file path as JSON and dump into dumpFilePath
# rdd = sc.parallelize(res)
# SQLContext.createDataFrame(rdd).collect()
# df = SQLContext.createDataFrame(rdd,['Order','CLass','Name', 'Value'])
#tempDumpFilePath = dumpFilePath + "/part-00000"
#if os.path.exists(tempDumpFilePath):
# os.remove(tempDumpFilePath)
#df.toJSON().saveAsTextFile(hdfsFilePath)
#tmpHdfsFilePath = hdfsFilePath + "/part-00000"
#subprocess.call(["hadoop","fs","-copyToLocal", tmpHdfsFilePath, dumpFilePath])
# Save and load model
#clusters.save(sc, "myModel")
#sameModel = KMeansModel.load(sc, "myModel")
training, test = data.randomSplit([0.85, 0.15], seed=11L)
training.cache()
# Run training algorithm to build the model
model = LogisticRegressionWithLBFGS.train(training, numClasses=39)
# Compute raw scores on the test set
predictionAndLabels = test.map(lambda lp: (float(model.predict(lp.features)), lp.label))
# Instantiate metrics object
metrics = MulticlassMetrics(predictionAndLabels)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
#accuracy = metrics.accuracy
accuracy = 1.0 * predictionAndLabels.filter(lambda (x, v): x == v).count() / test.count()
# print("Summary Stats")
# print("Precision = %s" % precision)
# print("Recall = %s" % recall)
# print("F1 Score = %s" % f1Score)
# print("Accuracy = %s" % accuracy)
# Statistics by class
labels = data.map(lambda lp: lp.label).distinct().collect()
# for label in sorted(labels):
# print("Class %s precision = %s" % (label, metrics.precision(label)))
# print("Class %s recall = %s" % (label, metrics.recall(label)))
# print("Class %s F1 Measure = %s" % (label, metrics.fMeasure(label, beta=1.0)))
# Weighted stats
def test_prfs():
# TODO: revised so that it will take user's inputs instead of hardcoded values
"""
Test Precision, Recall, Fscore, and Support on multiclass classification data
Input data: https://github.com/apache/spark/blob/master/data/mllib/sample_multiclass_classification_data.txt.
"""
# load the schemas (if existed)
# create a hdfs directory
#os.system("hdfs dfs -mkdir datasets")
# load the data file into the hdfs directory
os.system("hdfs dfs -put sample_multiclass_classification_data.txt datasets/sample_multiclass_classification_data.txt")
data = MLUtils.loadLibSVMFile(scsingleton.sc, "hdfs://localhost:9000/datasets/sample_multiclass_classification_data.txt")
# print data.take(1)
# ie. [LabeledPoint(1.0, (4,[0,1,2,3],[-0.222222,0.5,-0.762712,-0.833333]))]
# [ ( finalClassification, (numLabels, [label0, label1, label2, ..., labelN], [prob0, prob1, prob2, ..., probN]) ) ]
# split data into train (60%), test (40%)
trainingRDD, testRDD = data.randomSplit([0.6, 0.4])
trainingRDD.cache()
testRDD.cache()
with Timer() as t:
numTest = testRDD.count()
print "testRDD.count(): %s seconds" % t.secs
# run training algorithm to build the model
# without validation
with Timer() as t:
model = LogisticRegressionWithLBFGS.train(trainingRDD, numClasses=3)
print "LogisticRegressionWithLBFGS.train(trainingRDD, numClasses=3): %s seconds" % t.secs
# make a prediction
with Timer() as t:
testPredAndLabel = testRDD.map(lambda lp: (float(model.predict(lp.features)), lp.label))
print "testPredAndLabel: %s seconds" % t.secs
# calculate Precision, Recall, F1-score
metrics = MulticlassMetrics(testPredAndLabel)
print( "precision = %s" % metrics.precision() )
print( "recall = %s" % metrics.recall() )
print( "f1-score = %s" % metrics.fMeasure() )
# statistics by class
labels = data.map(lambda lp: lp.label).distinct().collect()
for label in sorted(labels):
print( "Class %s precision = %s" % (label, metrics.precision(label)) )
print( "Class %s recall = %s" % (label, metrics.recall(label)) )
print( "Class %s f1-score = %s" % (label, metrics.fMeasure(label, beta=1.0)) )
# weighted stats
print( "Weighted precision = %s" % metrics.weightedPrecision )
print( "Weighted recall = %s" % metrics.weightedRecall )
print( "Weighted f1-score = %s" % metrics.weightedFMeasure() )
print( "Weighted f(0.5)-score = %s" % metrics.weightedFMeasure(beta=0.5) )
print( "Weighted false positive rate = %s" % metrics.weightedFalsePositiveRate )
return
# Load testing data in LIBSVM format
#testdata = MLUtils.loadLibSVMFile(sc, loadTestingFilePath)
# Run training algorithm to build the model
model = LogisticRegressionWithLBFGS.train(traindata, numClasses=3)
# Compute raw scores on the test set
predictionAndLabels = testdata.map(lambda lp: (float(model.predict(lp.features)), lp.label))
# Instantiate metrics object
metrics = MulticlassMetrics(predictionAndLabels)
# Overall statistics
precision = metrics.precision()
recall = metrics.recall()
f1Score = metrics.fMeasure()
#confusion_matrix = metrics.confusionMatrix().toArray()
print("Summary Stats")
print("Precision = %s" % precision)
print("Recall = %s" % recall)
print("F1 Score = %s" % f1Score)
# Statistics by class
labels = traindata.map(lambda lp: lp.label).distinct().collect()
for label in sorted(labels):
print("Class %s precision = %s" % (label, metrics.precision(label)))
print("Class %s recall = %s" % (label, metrics.recall(label)))
print("Class %s F1 Measure = %s" % (label, metrics.fMeasure(label, beta=1.0)))