def testPerplexityOnRealData(self):
dtype = np.float64 # DTYPE
rd.seed(0xBADB055)
data = DataSet.from_files(words_file=AclWordPath, links_file=AclCitePath)
with open(AclDictPath, "rb") as f:
d = pkl.load(f)
data.convert_to_dtype(np.int32)
data.prune_and_shuffle(min_doc_len=50, min_link_count=2)
data.convert_to_undirected_graph()
data.convert_to_binary_link_matrix()
# IDF frequency for when we print out the vocab later
freq = np.squeeze(np.asarray(data.words.sum(axis=0)))
scale = np.reciprocal(1 + freq)
# Initialise the model
K = 10
model = lda.newModelAtRandom(data, K, dtype=dtype)
queryState = lda.newQueryState(data, model)
trainPlan = lda.newTrainPlan(iterations=300, logFrequency=50, fastButInaccurate=False, debug=True)
# Train the model, and the immediately save the result to a file for subsequent inspection
model, query, (bndItrs, bndVals, bndLikes) = lda.train (data, model, queryState, trainPlan)
# with open(newModelFileFromModel(model), "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()
vocab = lda.wordDists(model)
plt.imshow(vocab, interpolation="none", cmap = cm.Greys_r)
plt.show()
# Print out the most likely topic words
topWordCount = 10
kTopWordInds = [self.topWordInds(vocab[k,:], topWordCount) for k in range(K)]
like = lda.log_likelihood(data, model, query)
perp = perplexity_from_like(like, data.word_count)
print ("Prior %s" % (str(model.topicPrior)))
print ("Perplexity: %f\n\n" % perp)
for k in range(model.K):
print ("\nTopic %d\n=============================" % k)
print ("\n".join("%-20s\t%0.4f" % (d[kTopWordInds[k][c]], vocab[k][kTopWordInds[k][c]]) for c in range(topWordCount)))
def testCrossValPerplexityOnRealDataWithLdaGibbsInc(self):
ActiveFolds = 3
dtype = np.float64 # DTYPE
rd.seed(0xBADB055)
data = DataSet.from_files(words_file=AclWordPath, links_file=AclCitePath)
data.convert_to_dtype(np.int32) # Gibbs expects integers as input, regardless of model dtype
data.prune_and_shuffle(min_doc_len=MinDocLen, min_link_count=MinLinkCount)
# Training setup
TrainSamplesPerTopic = 10
QuerySamplesPerTopic = 2
Thin = 2
Debug = False
# Start running experiments
topicCounts = [5, 10, 15, 20, 25, 30, 35, 40, 45, 50]
for K in topicCounts:
trainPlan = lda_gibbs.newTrainPlan(K * TrainSamplesPerTopic, thin=Thin, debug=Debug)
queryPlan = lda_gibbs.newTrainPlan(K * QuerySamplesPerTopic, thin=Thin, debug=Debug)
trainPerps = []
queryPerps = []
for fold in range(ActiveFolds): # range(NumFolds):
trainData, queryData = data.cross_valid_split(fold, NumFolds)
estData, evalData = queryData.doc_completion_split()
model = lda_gibbs.newModelAtRandom(trainData, K, dtype=dtype)
query = lda_gibbs.newQueryState(trainData, model)
# Train the model, and the immediately save the result to a file for subsequent inspection
model, trainResult, (_, _, _) = lda_gibbs.train (trainData, model, query, trainPlan)
like = lda_gibbs.log_likelihood(trainData, model, trainResult)
perp = perplexity_from_like(like, trainData.word_count)
trainPerps.append(perp)
query = lda_gibbs.newQueryState(estData, model)
_, queryResult = lda_gibbs.query(estData, model, query, queryPlan)
like = lda_gibbs.log_likelihood(evalData, model, queryResult)
perp = perplexity_from_like(like, evalData.word_count)
queryPerps.append(perp)
trainPerps.append(sum(trainPerps) / ActiveFolds)
queryPerps.append(sum(queryPerps) / ActiveFolds)
print("K=%d,Segment=Train,%s" % (K, ",".join([str(p) for p in trainPerps])))
print("K=%d,Segment=Query,%s" % (K, ",".join([str(p) for p in queryPerps])))