def plotManifolds(all_train_x, all_train_y, all_test_x, trainSampleWeights, testSampleWeights, n_jobs=23):
mult = 20
trainSampleWeights = mult*(trainSampleWeights - trainSampleWeights.min()) + 2
testSampleWeights = mult*(testSampleWeights - testSampleWeights.min()) + 2
# ------ transform and plot ------
n_neighbors = 10
names = ['LLE']
transformers = [('LLE', LocallyLinearEmbedding(n_neighbors=n_neighbors, method='standard')),
('MDS', MDS(n_jobs=n_jobs)),
('Isomap', Isomap(n_neighbors=n_neighbors)),
('Spectral Embedding', SpectralEmbedding(n_neighbors=n_neighbors))]
fig = pylab.figure()
for i, (name, transformer) in enumerate(transformers):
print '---', name, '---'
# transform
t0 = time()
traindata = deepcopy(transformer).fit_transform(all_train_x)
testdata = deepcopy(transformer).fit_transform(all_test_x)
printDoneTime(t0)
# plot
ax = fig.add_subplot(2, len(transformers),2*i+1)
for surv in [0,1]:
color = "red" if surv==1 else "blue"
label = 'survived' if surv==1 else 'died'
ind = all_train_y==surv
pylab.scatter(traindata[ind,0], traindata[ind,1], c=color, alpha=0.7, s=trainSampleWeights, linewidths=0.5, label=label)
pylab.title(name + " (Training Data)")
ax.xaxis.set_major_formatter(NullFormatter())
ax.yaxis.set_major_formatter(NullFormatter())
ax.legend(loc="best")
pylab.axis('tight')
ax = fig.add_subplot(2, len(transformers),2*i+2)
pylab.scatter(testdata[:,0], testdata[:,1], c="grey", s=testSampleWeights)
pylab.title(name + " (Test Data)")
ax.xaxis.set_major_formatter(NullFormatter())
ax.yaxis.set_major_formatter(NullFormatter())
pylab.axis('tight')
pylab.show()
# trainX, trainY, testX, testY = loadObject("H:/allcvdata")[1]
# params = getParamsFromIndices([2, 2, 1, 0, 2], allParamsDict)
# pipe.set_params(**params)
# pipe.fit(trainX, trainY)
# print 'here'
# print accuracy_score(testY, pipe.predict(testX))
t0 = time()
# newpipe, best_params, score = fitClfWithGridSearch(name, pipe, allParamsDict, data, os.path.join(rootdir, 'intermediate results'),
# n_jobs=20, cvSplitNum=10, random_state=random_state, useJJ=useJJ, verbosity=2,
#
# maxLearningSteps=30, numConvergenceSteps=4, eliteProportion=0.1, saveCache=True,
# parentsProportion=0.4, populationSize=15,
# mutationProbability=0.3, mutationProportion=0.2, mutationStdDev=None, maxDuplicateProportion=0)
# print 'SCORE =', score
# print 'x'*50
buildModel(data, testData, fieldMaps, selectedClfs=['svc'], useJJ=useJJ, n_jobs=20, writeResults=True,
colNames=['sex', 'name', 'embarked'], cvNumSplits=10, random_states=random_states, verbose=True,
maxLearningSteps=30, numConvergenceSteps=5, eliteProportion=0.1, saveCache=True,
parentsProportion=0.4, populationSize=12, verbosity=2,
mutationProbability=0.3, mutationProportion=0.2, mutationStdDev=None, maxDuplicateProportion=0)
printDoneTime(t0)
print '>>>> FIN <<<<'