def plotClusterOverlapByLocationCheckins(place):
dataX, dataY = [], []
for location in locationClusterMeansIterator(place):
locationOverlaps = []
if len(location['clusters'])>=2:
for cluster1, cluster2 in combinations(location['clusters'], 2):
mu1, mu2, sd1, sd2 = cluster1[1], cluster2[1], cluster1[2], cluster2[2]
locationOverlaps.append(getWeitzmanOVL(mu1, mu2, sd1, sd2)[0])
dataY.append(np.mean(locationOverlaps))
dataX.append(location['details']['noOfCheckins'])
plt.scatter(dataX, dataY)
plt.show()
def plotOverlapDistribution(place):
oVLValues = []
for location in locationClusterMeansIterator(place):
locationOVLValues = []
if len(location['clusters'])>=2:
for cluster1, cluster2 in combinations(location['clusters'], 2):
mu1, mu2, sd1, sd2 = cluster1[1], cluster2[1], cluster1[2], cluster2[2]
ovl = getWeitzmanOVL(mu1, mu2, sd1, sd2)
locationOVLValues.append(ovl[0])
if locationOVLValues:
mean = np.mean(locationOVLValues)
if mean>0: oVLValues.append(mean)
plt.hist(oVLValues, 100, facecolor='green', alpha=0.75)
plt.show()
def iterateLocationsByOVLAndClustersType(place, type):
dataX, dataY = [], []
for location in locationClusterMeansIterator(place):
locationOVLValues = []
if len(location['clusters'])>=2:
for cluster1, cluster2 in combinations(location['clusters'], 2):
mu1, mu2, sd1, sd2 = cluster1[1], cluster2[1], cluster1[2], cluster2[2]
ovl = getWeitzmanOVL(mu1, mu2, sd1, sd2)
locationOVLValues.append(ovl[0])
if locationOVLValues:
lengthOfCluster = len(location['clusters'])
mean = np.mean(locationOVLValues)
locationType = None
if lengthOfCluster<=place['lowClusters'] and mean<=place['lowOVL']: locationType = CLUSTERS_OVL_TYPE_LOW_LOW
elif lengthOfCluster<=place['lowClusters'] and mean>=place['highOVL']: locationType = CLUSTERS_OVL_TYPE_LOW_HIGH
elif lengthOfCluster>=place['highClusters'] and mean<=place['lowOVL']: locationType = CLUSTERS_OVL_TYPE_HIGH_LOW
elif lengthOfCluster>=place['highClusters'] and mean>=place['highOVL']: locationType = CLUSTERS_OVL_TYPE_HIGH_HIGH
if locationType==type: yield location
def plotClusterOverlapInLocations(place):
dataX, dataY = [], []
for location in locationClusterMeansIterator(place):
locationOVLValues = []
if len(location['clusters'])>=2:
for cluster1, cluster2 in combinations(location['clusters'], 2):
mu1, mu2, sd1, sd2 = cluster1[1], cluster2[1], cluster1[2], cluster2[2]
ovl = getWeitzmanOVL(mu1, mu2, sd1, sd2)
locationOVLValues.append(ovl[0])
if locationOVLValues:
lengthOfCluster = len(location['clusters'])
mean = np.mean(locationOVLValues)
if lengthOfCluster<=place['lowClusters'] and mean<=place['lowOVL']: plt.scatter([lengthOfCluster], [mean], color='g')
elif lengthOfCluster<=place['lowClusters'] and mean>=place['highOVL']: plt.scatter([lengthOfCluster], [mean], color='y')
elif lengthOfCluster>=place['highClusters'] and mean<=place['lowOVL']: plt.scatter([lengthOfCluster], [mean], color='r')
elif lengthOfCluster>=place['highClusters'] and mean>=place['highOVL']: plt.scatter([lengthOfCluster], [mean], color='m')
else: plt.scatter([lengthOfCluster], [np.mean(locationOVLValues)], color='b')
plt.show()
