#%%
print 'running PCA without rank'
for kindex, key in enumerate(keyList):
resultDict[key]['PCA'] = dr.runPCANormal(dataSets[key], pars)
# overview of PCA results and weights
mp.plotPCAresults(dataSets, resultDict, keyList, pars, flag='PCA')
# correlate PCA with all sorts of stuff
mp.plotPCAcorrelates(dataSets, resultDict, keyList, pars, flag='PCA')
plt.show()
# plot 3D trajectory of PCA
#
#mp.plotPCAresults3D(dataSets, resultDict, keyList, pars, col = 'phase')
#plt.show()
mp.plotPCAresults3D(dataSets, resultDict, keyList, pars, col='velocity')
mp.plotPCAresults3D(dataSets, resultDict, keyList, pars, col='turns')
#plt.show()
# save the most prominent 3 neurons for each PCA axis
pars['useRank'] = 1
print 'running PCA with rank'
for kindex, key in enumerate(keyList):
resultDict[key]['RankPCA'] = dr.runPCANormal(dataSets[key], pars)
# overview of PCA results and weights
mp.plotPCAresults(dataSets, resultDict, keyList, pars, flag='RankPCA')
mp.plotPCAcorrelates(dataSets, resultDict, keyList, pars, flag='RankPCA')
plt.show()
# plot 3D trajectory of PCA
mp.averageResultsPCA(res, keys, labelsPCA, colorsPCA, fitmethod="PCA")
plt.show()
###############################################
#
# use svm to predict discrete behaviors
#
##############################################
if svm:
# overview of SVM results and weights
mp.plotPCAresults(dataSets, resultDict, keyList, pars, flag='SVM')
plt.show()
# plot 3D trajectory of SVM
mp.plotPCAresults3D(dataSets,
resultDict,
keyList,
pars,
col='etho',
flag='SVM')
plt.show()
# mp.plotPCAresults3D(dataSets, resultDict, keyList, pars, col = 'time', flag = 'SVM')
# plt.show()
# mp.plotPCAresults3D(dataSets, resultDict, keyList, pars, col = 'velocity', flag = 'SVM')
# plt.show()
# mp.plotPCAresults3D(dataSets, resultDict, keyList, pars, col = 'turns', flag = 'SVM')
# plt.show()
###############################################
#
# run PCA and store results
#
##############################################
# run PCA and store results
#
##############################################
#%%
if pca:
print 'running PCA'
for kindex, key in enumerate(keyList):
resultDict[key]['PCA'] = dr.runPCANormal(dataSets[key],
pars,
whichPC=0)
# overview of data ordered by PCA
mp.plotDataOverview2(dataSets, keyList, resultDict)
# overview of PCA results and weights
mp.plotPCAresults(dataSets, resultDict, keyList, pars)
plt.show()
mp.plotPCAresults3D(dataSets, resultDict, keyList, pars, col='time')
plt.show()
# color by before and after
colorBy = np.zeros(dataSets[key]['Neurons']['Activity'].shape[1])
colorBy[:int(dataSets[key]['Neurons']['Activity'].shape[1] / 2.)] = 1
mp.plotPCAresults3D(dataSets,
resultDict,
keyList,
pars,
col='Immobilization',
colorBy=colorBy)
plt.show()
plt.show(block=True)
resultDict['PCA'] = {}
resultDict['PCA2'] = {}
testset=None)
###############################################
#
# use behavior triggered averaging to create non-othogonal axes
#
##############################################
if bta:
for kindex, key in enumerate(keyList):
print 'running BTA'
resultDict[key]['BTA'] = dr.runBehaviorTriggeredAverage(
dataSets[key], pars)
mp.plotPCAresults(dataSets, resultDict, keyList, pars, flag='BTA')
plt.show()
mp.plotPCAresults3D(dataSets,
resultDict,
keyList,
pars,
col='etho',
flag='BTA')
plt.show()
###############################################
#
# use svm to predict discrete behaviors
#
##############################################
if svm:
for kindex, key in enumerate(keyList):
print 'running SVM'
splits = resultDict[key]['Training']
resultDict[key]['SVM'] = dr.discreteBehaviorPrediction(
dataSets[key], pars, splits)