def predict(M_c, X_L, X_D, Y, Q, n, get_next_seed, return_samples=False):
# Predict is currently the same as impute except that the row Id in the query must lie outside the
# length of the table used to generate the model
# For now, we will just call "impute" and leave it to the user to generate the query correctly
# FIXME: allow more than one cell to be predicted
assert(len(Q)==1)
if return_samples:
e, samples = su.impute(M_c, X_L, X_D, Y, Q, n, get_next_seed, return_samples=True)
else:
e = su.impute(M_c, X_L, X_D, Y, Q, n, get_next_seed)
return e
def impute(self, M_c, X_L, X_D, Y, Q, seed, n):
"""Impute values from the predictive distribution of the given latent state
:param seed: The random seed
:type seed: int
:param M_c: The column metadata
:type M_c: dict
:param X_L: the latent variables associated with the latent state
:type X_L: dict
:param X_D: the particular cluster assignments of each row in each view
:type X_D: list of lists
:param Y: A list of constraints to apply when sampling. Each constraint
is a triplet of (r,d,v): r is the row index, d is the column
index and v is the value of the constraint
:type Y: list of lists
:param Q: A list of values to sample. Each value is doublet of (r, d):
r is the row index, d is the column index
:type Q: list of lists
:param n: the number of samples to use in the imputation
:type n: int
:returns: list of floats -- imputed values in the same order as
specified by Q
"""
get_next_seed = make_get_next_seed(seed)
e = su.impute(M_c, X_L, X_D, Y, Q, n, get_next_seed)
return e
def predict(M_c, X_L, X_D, Y, Q, n, get_next_seed, return_samples=False):
# Predict is currently the same as impute except that the row Id in the query must lie outside the
# length of the table used to generate the model
# For now, we will just call "impute" and leave it to the user to generate the query correctly
# FIXME: allow more than one cell to be predicted
assert (len(Q) == 1)
if return_samples:
e, samples = su.impute(M_c,
X_L,
X_D,
Y,
Q,
n,
get_next_seed,
return_samples=True)
else:
e = su.impute(M_c, X_L, X_D, Y, Q, n, get_next_seed)
return e
def impute(self, M_c, X_L, X_D, Y, Q, seed, n):
"""Impute values from predictive distribution of the given latent state.
:param Y: A list of constraints to apply when sampling. Each constraint
is a triplet of (r,d,v): r is the row index, d is the column
index and v is the value of the constraint
:type Y: list of lists
:param Q: A list of values to sample. Each value is doublet of (r, d):
r is the row index, d is the column index
:type Q: list of lists
:param n: the number of samples to use in the imputation
:type n: int
:returns: list of floats -- imputed values in the same order as
specified by Q
"""
get_next_seed = make_get_next_seed(seed)
e = su.impute(M_c, X_L, X_D, Y, Q, n, get_next_seed)
return e
def impute_table(T, M_c, X_L_list, X_D_list, numDraws, get_next_seed):
T_imputed = copy(T)
num_rows = len(T)
num_cols = len(T[0])
# Identify column types
col_names = numpy.array(
[M_c['idx_to_name'][str(col_idx)] for col_idx in range(num_cols)])
coltype = []
for colindx in range(len(col_names)):
if M_c['column_metadata'][colindx][
'modeltype'] == 'normal_inverse_gamma':
coltype.append('continuous')
else:
coltype.append('multinomial')
rowsWithNans = [i for i in range(len(T)) if any(isnan_mixedtype(T[i]))]
print rowsWithNans
Q = []
for x in rowsWithNans:
y = [y for y in range(len(T[0])) if isnan_mixedtype([T[x][y]])]
Q.extend(zip([x] * len(y), y))
numImputations = len(Q)
# Impute missing values in table
values_list = []
for queryindx in range(len(Q)):
values = su.impute(M_c, X_L_list, X_D_list, [], [Q[queryindx]],
numDraws, get_next_seed)
values_list.append(values)
# Put the samples back into the data table
for imputeindx in range(numImputations):
imputed_value = values_list[imputeindx]
if coltype[Q[imputeindx][1]] == 'multinomial':
imputed_value = M_c['column_metadata'][
Q[imputeindx][1]]['value_to_code'][imputed_value]
T_imputed[Q[imputeindx][0]][Q[imputeindx][1]] = imputed_value
return T_imputed
def impute_table(T, M_c, X_L_list, X_D_list, numDraws, get_next_seed):
T_imputed = copy(T)
num_rows = len(T)
num_cols = len(T[0])
# Identify column types
col_names = numpy.array([M_c['idx_to_name'][str(col_idx)] for col_idx in range(num_cols)])
coltype = []
for colindx in range(len(col_names)):
if M_c['column_metadata'][colindx]['modeltype'] == 'normal_inverse_gamma':
coltype.append('continuous')
else:
coltype.append('multinomial')
rowsWithNans = [i for i in range(len(T)) if any(isnan_mixedtype(T[i]))]
print(rowsWithNans)
Q = []
for x in rowsWithNans:
y = [y for y in range(len(T[0])) if isnan_mixedtype([T[x][y]])]
Q.extend(zip([x]*len(y), y))
numImputations = len(Q)
# Impute missing values in table
values_list = []
for queryindx in range(len(Q)):
values = su.impute(M_c, X_L_list, X_D_list, [], [Q[queryindx]], numDraws, get_next_seed)
values_list.append(values)
# Put the samples back into the data table
for imputeindx in range(numImputations):
imputed_value = values_list[imputeindx]
if coltype[Q[imputeindx][1]] == 'multinomial':
imputed_value = M_c['column_metadata'][Q[imputeindx][1]]['value_to_code'][imputed_value]
T_imputed[Q[imputeindx][0]][Q[imputeindx][1]] = imputed_value
return T_imputed
