def main():
data = extract_data_from_csv("")
derivatives = computeDerivatives(data)
"""sns.set(color_codes=True)
sns.distplot(derivatives[4])
plt.show()"""
intervals = computeConfidenceIntervals(derivatives)
print(intervals)
print(findSuspicious(derivatives, intervals))
def main():
data = extract_data_from_csv("")
predictor = Predictor(0, 2, 10000, distance_in_future=15)
vectors, labels = predictor.generateVectors(data)
predicted_temp = predictor.generatePredictions_oneValue(
get_classifier(""), vectors)[-1]
print("Predicted %d degrees Celcius (+/- 2 degrees) in 15 minutes" %
predicted_temp)
if predicted_temp > CUTOFF:
print("Coffee should be made!")
sendEmail(
"Our predictive models have determined that your machine should be turned on within the next 15 minutes.",
"Coffee machine should be turned on!")
def generateCostDistribution(self,falsified,size,num_datapoints_testing = -1):
training_proportion = 0.5
print("Gathering data")
raw_data = extract_data_from_csv("")
training_data = raw_data[0:int(len(raw_data)*training_proportion)]
training_vectors,training_labels = self.generateVectors(training_data)
print("Finished gathering data")
print("training model")
clf = getTrainedModel(training_vectors,training_labels)
print("model trained")
costs = list()
for i in range(size):
testing_data = getRandomTestingDataSample(0.03,training_proportion,raw_data,num_datapoints_testing)
costs.append(self.computeCost(falsified,clf,testing_data)[0])
return costs
def quickComparison(self):
print("Gathering data")
raw_data = extract_data_from_csv("")
training_data, testing_data = separateData(raw_data,0.75,0,1)
training_vectors,training_labels = self.generateVectors(training_data)
print("Sample vector: "+str(training_vectors[0]))
print("Finished gathering data")
print("training model")
clf = getTrainedModel(training_vectors,training_labels)
print("model trained")
cost_normal,testing_labels,predictions = self.computeCost(False,clf,testing_data)
cost_falsified,testing_labels_falsified,predictions_falsified = self.computeCost(True,clf,testing_data)
#incremental_costs = calculate_costs(30,predictions,testing_labels)
#plt.figure(1)
#plt.plot(incremental_costs)
#plt.plot([val/30 for val in testing_labels[::30]])
#plt.show()
print("Cost when data is not falsified: "+str(cost_normal))
print("Cost when data is falsified: "+str(cost_falsified))
return predictions,predictions_falsified,testing_labels
def get_last_window(num_minutes, root):
data = extract_data_from_csv(root)
data = data[(len(data) - num_minutes):(len(data))]
return data