data = np.zeros((D, model.V),dtype=int)
for d in xrange(D):
doc = model.generate(N=N, keep=True)
data[d,:] = doc.w
# def test_geweke_lda():
if __name__ == "__main__":
N_iter = 50000
T = 3 # Number of topics
D = 10 # Number of documents
V = 10 # Number of words
N = 20 # Number of words per document
alpha_beta = 1.0
# Sample a GP
model = StickbreakingCorrelatedLDA(T, V, alpha_beta=alpha_beta)
# Run a Geweke test
thetas = []
betas = []
for itr in progprint_xrange(N_iter):
# Resample the data
resample_data(model, D, N)
# Resample the parameters
model.resample_model()
# Update our samples
thetas.append(model.thetas.copy())
betas.append(model.beta.copy())
## DEBUG! Set beta and theta to true values
init_to_true = False
std_model = StandardLDA(T,V,alpha_beta,alpha_theta)
std_model.beta = true_lda.beta if init_to_true else std_model.beta
std_results = \
train(std_model, thetas=true_lda.thetas if init_to_true else None)
std_collapsed_model = StandardLDA(T,V,alpha_beta,alpha_theta)
std_collapsed_model.beta = true_lda.beta if init_to_true else std_collapsed_model.beta
std_collapsed_results = \
train(std_collapsed_model,
method='resample_model_collapsed',
thetas=true_lda.thetas if init_to_true else None)
sb_model = StickbreakingCorrelatedLDA(T, V, alpha_beta)
sb_model.beta = true_lda.beta if init_to_true else sb_model.beta
sb_results = \
train(sb_model, thetas=true_lda.thetas if init_to_true else None)
ln_model = LogisticNormalCorrelatedLDA(T, V, alpha_beta)
ln_model.beta = true_lda.beta if init_to_true else ln_model.beta
ln_results = \
train(ln_model, thetas=true_lda.thetas if init_to_true else None)
all_results = [sb_results, ln_results, std_results, std_collapsed_results]
all_labels = ["SB Corr. LDA", "LN Corr. LDA", "Std. LDA", "Collapsed LDA"]
# all_results = [std_results, std_collapsed_results]
# all_labels = ["Std. LDA", "Collapsed LDA"]
# all_results = [ln_results]
# all_labels = ["LN Corr. LDA"]
# def test_geweke_lda():
if __name__ == "__main__":
N_iter = 5000
T = 3 # Number of topics
D = 10 # Number of documents
V = 20 # Number of words
N = 20 # Number of words per document
alpha_beta = 1.0
# Generate synthetic data
data = np.random.poisson(2, (D,V))
data = csr_matrix(data)
# Sample a GP
model = StickbreakingCorrelatedLDA(data, T, alpha_beta=alpha_beta)
# Run a Geweke test
thetas = []
betas = []
for itr in progprint_xrange(N_iter):
# Resample the data
model.generate(N, keep=True)
# Resample the parameters
model.resample()
# Update our samples
thetas.append(model.theta.copy())
betas.append(model.beta.copy())
data = np.zeros((D, model.V),dtype=int)
for d in xrange(D):
doc = model.generate(N=N, keep=True)
data[d,:] = doc.w
# def test_geweke_lda():
if __name__ == "__main__":
N_iter = 50000
T = 3 # Number of topics
D = 10 # Number of documents
V = 10 # Number of words
N = 20 # Number of words per document
alpha_beta = 1.0
# Sample a GP
model = StickbreakingCorrelatedLDA(T, V, alpha_beta=alpha_beta)
# Run a Geweke test
thetas = []
betas = []
for itr in progprint_xrange(N_iter):
# Resample the data
resample_data(model, D, N)
# Resample the parameters
model.resample_model()
# Update our samples
thetas.append(model.thetas.copy())
betas.append(model.beta.copy())