supplementalVocabularyWords=['call', 'to', 'your'])
# Apply the featurerizer to the raw data sets to produce feature vectors. In this case, each message will be converted to an array
# with one bit per feature that is 1 if the message has the feature, and 0 if the message does not have the feature.
xTrain = featurizer.Featurize(xTrainRaw)
xValidate = featurizer.Featurize(xValidateRaw)
xTest = featurizer.Featurize(xTestRaw)
print("\n - Inspect the features -")
for i in range(len(xTrain[0])):
print(featurizer.GetFeatureInfo(i))
print("\n - Inspect feature values for a few training samples -")
for i in range(5):
print(yTrain[i], "-", xTrain[i], xTrainRaw[i])
# Now let's up our modeling game (as compared to predicting the most common class)
# we'll use a heuristic (hand-tuned) linear model.
import MachineLearningCourse.MLUtilities.Learners.LinearHeuristicModel as LinearHeuristicModel
model = LinearHeuristicModel.LinearHeuristicModel()
model.fit(xTrain, yTrain, -1.0, [.75, .75, .75, .25, .25])
print("\n - Inspect the weights on the heuristically-tuned model -")
model.visualize()
yValidatePredicted = model.predict(xValidate)
import MachineLearningCourse.MLUtilities.Evaluations.EvaluateBinaryClassification as EvaluateBinaryClassification
EvaluateBinaryClassification.ExecuteAll(yValidate, yValidatePredicted)
# Learn the logistic regression model
print("Learning the logistic regression model:")
import MachineLearningCourse.MLUtilities.Learners.LogisticRegression as LogisticRegression
logisticRegressionModel = LogisticRegression.LogisticRegression()
logisticRegressionModel.fit(xTrain,
yTrain,
stepSize=1.0,
convergence=0.005)
#############################
# Evaluate the model
import MachineLearningCourse.MLUtilities.Evaluations.EvaluateBinaryClassification as EvaluateBinaryClassification
print("\nLogistic regression model:")
logisticRegressionModel.visualize()
EvaluateBinaryClassification.ExecuteAll(
yValidate,
logisticRegressionModel.predict(xValidate,
classificationThreshold=0.5))
#################
# You may find the following module helpful for making charts. You'll have to install matplotlib (see the lecture notes).
#
# import MachineLearningCourse.MLUtilities.Visualizations.Charting as Charting
#
# # trainLosses, validationLosses, and lossXLabels are parallel arrays with the losses you want to plot at the specified x coordinates
#
# Charting.PlotSeries([trainLosses, validationLosses], ['Train', 'Validate'], lossXLabels, chartTitle="Logistic Regression", xAxisTitle="Gradient Descent Steps", yAxisTitle="Avg. Loss", outputDirectory=kOutputDirectory, fileName="3-Logistic Regression Train vs Validate loss")