pymc.pytensorf.convert_data#

pymc.pytensorf.convert_data(data)[source]#

Convert user data to an array or PyTensor variable with a PyMC dtype.

Parameters:
dataarray_like, pandas object, scipy sparse matrix or Variable

Data to convert. Nulls in pandas objects and NaNs in numpy arrays become masked entries. Variables and sparse matrices are never masked.

Returns:
convertedndarray, MaskedArray, sparse matrix or Variable

data cast to floatX if it has no dtype. Otherwise floats go to floatX, integers to the matching intX (int32 when floatX is float64), and other dtypes are unchanged. A masked array with nothing masked comes back filled.

Examples

import numpy as np
from pymc.pytensorf import convert_data

convert_data([1, 2, 3])  # floatX array
convert_data(np.array([1, 2, 3]))  # intX array
convert_data(np.array([1.0, np.nan]))  # masked array, second entry masked