Table

heavylight.Table

Table provides multi-key compatible high performance table lookup.

__init__(df: pd.DataFrame, rectify: Union[bool, None] = False, safe: Union[bool, None] = True)

Initialise a table from a dataframe.

Parameters:

Name	Type	Description	Default
df	DataFrame	the pandas dataframe used to initialise the table	required
rectify	Union[bool, None]	force table to be rectangular (default False)	False
safe	Union[bool, None]	validates that integers are between bounds (default True)	True

Tables should be in long format

• the final column containing the values to look up
• all other columns contain keys to lookup
• tables should be contingous, i.e. no gaps in integer keys.
• tables should be complete if viewed as square matrixes (i.e. all combinations of keys are input). If not, you should fill any gaps with np.nan or a suitable value.

The type of key is determined by the suffix on the dataframe df column names: |int: integers (...0, 1, 2, 3...), can start and end anywhere, but must be consecutive |int_bound: as |int but any values are constrained to the lowest and highest values. |int_cat: as |str, categorical integers. |str': keys are interpreted as strings, e.g. 'M' and 'F'|band: key is numeric and treated as the upper bound on a lookup.|float`: not currently available due to floating point equality, use int or band depending on use case.

rectify(df: pd.DataFrame, fill=np.nan) -> pd.DataFrame staticmethod

Convert a triangular (incomplete) dataframe into a valid rectangular dataframe

any missing points will be filled with fill, default: np.nan

read_excel(spreadsheet_path, sheet_name) classmethod

Read in a table from an excel sheet, for more control pass in the dataframe using __init__

read_csv(csv_path) classmethod

Read in a table from an csv file, for more control pass in the dataframe using __init__