neurotools.util.pandas module

neurotools.util.pandas.get(df, col, val, drop=True)

Get all rows with column matching value (discards the column matched)

Parameters:

df (pandas.DataFrame)

Return type:

pd.DataFrame

neurotools.util.pandas.is_empty_col(c)

Check if a value is NaN

Parameters:

c (object) – tuple of str or nan representing multi- level column name.

Returns:

is_empty

Return type:

boolean

neurotools.util.pandas.from_hierarchical_columns(df, delimeter='__', deep=False)

Flatten hierarchical columns into strings. Inverse of to_hierarchical_columns.

Parameters:

• df (pandas.DataFrame)

• delimeter (str, default '__') – String delimiter for hierarchical column levels.

• deep (boolean, default False) – Whether retured array is a deep copy.

Return type:

pd.DataFrame

neurotools.util.pandas.to_ndarray(df, flatten_columns=True)

Convert DataFrame to structured np.ndarray.

Parameters:

• df (pandas.DataFrame)

• flatten_columns (boolean; default True) – Whether to flatten hierarchical columns by merging the coulmn-name tuples using the delimeter '__'.

Returns:

Dataframe as numpy structured array

Return type:

np.ndarray

neurotools.util.pandas.camelcase_columns(df)

Switch column names from Python’s underscore_convention to the Java/Matlab style lowerCaseCamelCase convention.

Your colleagues working in Matlab may appreciate this. This frees up underscore _ to use as a delimeter for hierarchical columns. Inverse of snakecase_columns(df).

Parameters:

df (pandas.DataFrame)

Return type:

pd.DataFrame

neurotools.util.pandas.snakecase_columns(df)

Switch column names from underscoreDelimited_camelCase to flattened-hierarical snake_case__columns. Inverse of camelcase_columns(df).

Parameters:

df (pandas.DataFrame)

Return type:

pd.DataFrame

neurotools.util.pandas.ndarray_from_mat(filename, key=None)

Load structured np.ndarray from matfile

Parameters:

• filename (str) – Matfile path to load

• flatten_columns (boolean; default True) – Whether to flatten hierarchical columns by merging the coulmn-name tuples using the delimeter '__'.

Returns:

• data (np.ndarray) – numpy structured array

• readme (str) – README string from this matfile, if present.

neurotools.util.pandas.read_mat(filename, key=None, restore_hierarchical_columns=True, delimeter='__')

Load DataFrame from matfile.

Parameters:

• filename (str) – Matfile path to load

• key (str; default None) – Name of variable to load from matfile. If none, expects matfile containing only a single variable.

• restore_hierarchical_columns (boolean; default True) – to_mat flattens hierarchical columns by merging the coulmn-name tuples using the delimeter '__'. If true, we will restore the hierarchical column structre in the loaded DataFrame.

• delimeter (str, default '__') – String delimiter for hierarchical column levels.

Return type:

pd.DataFrame

neurotools.util.pandas.to_mat(df, saveas, tablename='data', readme=None)

Save dataframe as .mat matfile.

Parameters:

• df (pandas.DataFrame)

• saveas (str) – File path to save to

• tablename (str; default 'data') – Variable name to use for dataframe in the matfile.

• readme (str; default None) – Optional string to save in a README variable to document the contents of this .mat archive.

neurotools.util.pandas.conform(x, dtype=<class 'numpy.float32'>, fill_value=nan)

Recursively coerce nested iterables into a numpy array with a uniform shape by padding uneven shapes with np.NaN or ~0.

Useful for ensuring columns have a uniform type signature.

Parameters:

• x (iterable) – Hierarichal (nested) iterable structure that we wish to coerce into a single np.ndarray

• dtype (np.dtype; default np.float32) – Desired type of the resulting array. Float and int types are supported. Float types will pad with np.NaN. Int types will pad with ~0.

• fill_value (number; default np.NaN)

Return type:

np.ndarray

neurotools.util.pandas.add_column_from_dict(df, coldata, default, name)

neurotools.util.pandas.dfhcat(*args)

neurotools.util.pandas.dfvcat(*args)

neurotools.util.pandas.check_column_signs(df)