Homec4science
Diffusion EpilepsyPerformanceMetricsLib (master) Browse

PerformanceMetricsLib/script_metricsDemo.py
73b6c2c475f4master

Last Change
Raw File

script_metricsDemo.py
View Options

''' Script to show how to use PerformanceMetricsLib '''
__author__ = "Una Pale"
__email__ = "una.pale at epfl.ch"
from PerformanceMetricsLib import *
## SETUP
# defining various parameters for postprocessing and performance measuring
class PerfParams:
samplFreq = 1 #in Hz, e.g. 1 sample=1sec
toleranceFP_befEvent = 1 #in sec
toleranceFP_aftEvent = 2 #in sec
eventStableLenToTest = 5 #in sec
eventStablePercToTest = 0.5
distanceBetween2events = 3 #in sec
bayesProbThresh =1.5 #bayes probability threshold (probably has to be tuned)
perfMetrics = EventsAndDurationPerformances(PerfParams)
############################################################################################
############################################################################################
## DIFFERENT TEST EXAMPLES
# trueLabels=np.array([0,0,0,0,0,0,0,0,1,1,1,1,1,1,0,0,0,0,0,0])
# predictions=np.array([0,0,0,0,0,0,0,0,1,1,1,1,1,1,0,0,0,0,0,0])
# predProbab=np.array([0.9,0.8,0.7,0.8,0.7,0.8,0.9,0.6,0.5,0.6,0.7,0.8,0.9,0.9,0.9,0.6,0.7,0.8,0.8,0.8])
# trueLabels=np.array([0,0,0,0,0,0,0,0,1,1,1,1,1,1,0,0,0,0,0,0])
# predictions=np.array([0,0,1,1,1,0,0,0,1,1,1,1,1,1,0,0,0,0,0,0])
# predProbab=np.array([0.9,0.8,0.7,0.8,0.7,0.8,0.9,0.6,0.5,0.6,0.7,0.8,0.9,0.9,0.9,0.6,0.7,0.8,0.8,0.8])
trueLabels=np.array([0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,0,0,0,0,0,0])
predictions=np.array([0,0,1,1,1,0,0,0,0,0,0,1,1,1,1,1,1,0,0,0,0,0,0])
predProbab=np.array([0.9,0.8,0.7,0.8,0.7,0.8,0.9,0.8,0.7,0.8,0.6,0.5,0.6,0.7,0.8,0.9,0.9,0.9,0.6,0.7,0.8,0.8,0.8])
############################################################################################
############################################################################################
## MEASURING PERFORMANCE
#all 9 performance metrics - on raw predictions, without smoothing
# firs three are on the level of events (sensitivity, precision, F1score)
# next three are on the level of duration (sensitivity, precision, F1score)
# then mean or F1E and F1D, geoMean of F1E and F1DE
# last is number of false positives that would be per day (linear interpolation from numFP in current sample)
performancesNoSmooth= perfMetrics.performance_all9(predictions, trueLabels)
print(performancesNoSmooth)
# performance after 2 types of postprocessing (moving average and bayes smoothing)
(performanceMetrics, smoothedPredictions) = perfMetrics.calculatePerformanceAfterVariousSmoothing(predictions, trueLabels,predProbab)
print(performanceMetrics)
#visualization of predictions after smoothing
allPredictSmoothing=np.vstack((trueLabels, predictions, smoothedPredictions['MovAvrg'], smoothedPredictions['MovAvrg&Merge'], smoothedPredictions['Bayes'], smoothedPredictions['Bayes&Merge']))
print(allPredictSmoothing)
c4science · Help