def testOnRealData(self):
rd.seed(0xDAFF0D12)
# path = "/Users/bryanfeeney/Desktop/NIPS"
# with open(path + "/ar.pkl", "rb") as f:
# X, W, _, dic = pkl.load(f)
path = "/Users/bryanfeeney/Desktop/SmallerDB-NoCJK-WithFeats-Fixed"
with open(path + "/all-in-one.pkl", "rb") as f:
W, X, dic = pkl.load(f)
if W.dtype != DTYPE:
W = W.astype(DTYPE)
if X.dtype != DTYPE:
X = X.astype(DTYPE)
D,T = W.shape
_,F = X.shape
freq = np.squeeze(np.asarray(W.sum(axis=0)))
scale = np.reciprocal(1. + freq)
K = 10
P = 5
model = stm.newModelAtRandom(X, W, P, K, 0.1, 0.1, dtype=DTYPE)
queryState = stm.newQueryState(W, model)
trainPlan = stm.newTrainPlan(iterations=50, logFrequency=1, debug=True)
model, query, (bndItrs, bndVals, bndLikes) = stm.train (W, X, model, queryState, trainPlan)
with open(newModelFile("stm-yv-bohn-nips-ar", K, None), "wb") as f:
pkl.dump ((model, query, (bndItrs, bndVals, bndLikes)), f)
# Plot the evolution of the bound during training.
fig, ax1 = plt.subplots()
ax1.plot(bndItrs, bndVals, 'b-')
ax1.set_xlabel('Iterations')
ax1.set_ylabel('Bound', color='b')
ax2 = ax1.twinx()
ax2.plot(bndItrs, bndLikes, 'r-')
ax2.set_ylabel('Likelihood', color='r')
fig.show()
plt.show()
# Print the top words
topWordCount = 100
kTopWordInds = [self.topWordInds(dic, model.vocab[k,:] * scale, topWordCount) \
for k in range(K)]
print ("Perplexity: %f\n\n" % ctm.perplexity(W, model, query))
print ("\t\t".join (["Topic " + str(k) for k in range(K)]))
print ("\n".join ("\t".join (dic[kTopWordInds[k][c]] + "\t%0.4f" % model.vocab[k][kTopWordInds[k][c]] for k in range(K)) for c in range(topWordCount)))
def _testLikelihoodOnModelDerivedExample(self):
print("Cross-validated likelihoods on model-derived example")
rd.seed(0xBADB055) # Global init for repeatable test
D, T, K, F, P = 200, 100, 10, 12, 8
tpcs, vocab, docLens, X, W = self._sampleFromModel()
plt.imshow(vocab, interpolation="none", cmap = cm.Greys_r)
plt.show()
W = W.astype(DTYPE)
X = X.astype(DTYPE)
# Create the cross-validation folds
folds = 5
foldSize = ceil(D / 5)
querySize = foldSize
trainSize = D - querySize
trainLikely = []
trainWordCount = []
queryLikely = []
queryWordCount = []
for fold in range(folds):
# Split the datasets
start = fold * foldSize
end = start + trainSize
trainSet = np.arange(start,end) % D
querySet = np.arange(end, end + querySize) % D
X_train, W_train = X[trainSet,:], W[trainSet,:]
X_query, W_query = X[querySet,:], W[querySet,:]
# Train the model
model = stm.newModelAtRandom(X_train, W_train, P, K, 0.1, 0.1, dtype=DTYPE)
queryState = stm.newQueryState(W_train, model)
plan = stm.newTrainPlan(iterations=100, logFrequency=1)
model, query, (bndItrs, bndVals, bndLikes) = stm.train (W_train, X_train, model, queryState, plan)
# Plot the evolution of the bound during training.
fig, ax1 = plt.subplots()
ax1.plot(bndItrs, bndVals, 'b-')
ax1.set_xlabel('Iterations')
ax1.set_ylabel('Bound', color='b')
ax2 = ax1.twinx()
ax2.plot(bndItrs, bndLikes, 'r-')
ax2.set_ylabel('Likelihood', color='r')
fig.show()
plt.show()
# Plot the topic covariance
self._plotCov(model)
# Plot the vocab
plt.imshow(model.vocab, interpolation="none", cmap = cm.Greys_r)
plt.show()
# Calculating the training set likelihood
trainLikely.append(stm.log_likelihood(W_train, model, queryState))
trainWordCount.append(W_train.data.sum())
# Now query the model.
plan = stm.newTrainPlan(iterations=100)
queryState = stm.newQueryState(W_query, model)
model, queryState = stm.query(W_query, X_query, model, queryState, plan)
queryLikely.append(stm.log_likelihood(W_query, model, queryState))
queryWordCount.append(W_query.data.sum())
for fold in range(folds):
trainPerp = np.exp(-trainLikely[fold]/trainWordCount[fold])
queryPerp = np.exp(-queryLikely[fold]/queryWordCount[fold])
print("Fold %3d: Train-set Likelihood: %12f \t Query-set Likelihood: %12f" % (fold, trainLikely[fold], queryLikely[fold]))
print(" Perplexity: %12.2f \t Perplexity: %12.2f" % (trainPerp, queryPerp))
self.assertTrue(queryPerp < 60.0) # Maximum perplexity.
self.assertTrue(trainPerp < 60.0)
print("End of Test")
