cascade.utils.tables#

Wrapper for .csv files.

__init__(csv_file_path: str, read_csv_kwargs=None, *args: Any, **kwargs: Any) → None[source]#

Passes all read_csv_kwargs to pd.read_csv

Parameters:

csv_file_path – path to the .csv file

Table dataset which allows to easily define and compute features

Example

import pandas as pd
from cascade.utils.tables import FeatureTable
df = pd.read_csv(r'data .csv', index_col=0)
df

# id  count  name
# 0   0      1   aaa
# 1   1      5   bbb
# 2   2      0   ccc

ft = FeatureTable(df)
ft.get_features()

# ['id', 'count', ' name']

ft.add_feature('square', lambda df: df['count'] * df['count'])

def counts(df):
    return df['count'] * 2, df['count'] * 3

ft.add_feature(('count_2', 'count_3'), counts)
ft.get_features()

# ['id', 'count', ' name', 'square', ('count_2', 'count_3')]

ft.get_table(['count', ('count_2', 'count_3')])

# count  count_2  count_3
# 0      1        2        3
# 1      5       10       15
# 2      0        0        0

__init__(table: Union[TableDataset, DataFrame], *args: Any, **kwargs: Any) → None[source]#

Parameters:

table (Union[TableDataset, pd.DataFrame]) – The table to wrap

add_feature(name: Union[str, Tuple[str]], func: Callable[[DataFrame], Union[Series, Tuple[str]]], *args: Any, **kwargs: Any) → None[source]#

Adds computable feature to the table in a lazy manner

Parameters:

• name (Union[str, Tuple[str]]) – Feature name

• func (Callable[[pd.DataFrame], Union[pd.Series, Tuple[str]]]) – Function that accepts pd.DataFrame and returns a new column

get_features() → List[Union[str, Tuple[str]]][source]#

Returns the list of feature names with all computed features added before

Returns:

List of feature names

Return type:

List[Union[str, Tuple[str]]]

get_meta() → List[Dict[Any, Any]][source]#

Returns:

meta – A list where last element is this dataset’s metadata. Meta can be anything that is worth to document about the dataset and its data. This is done in form of list to enable cascade-like calls in Modifiers and Samplers.

Return type:

Meta

get_table(features: Optional[Union[str, List[Union[str, Tuple[str]]]]] = None, dropna: bool = False) → DataFrame[source]#

Returns internal pd.DataFrame

Parameters:

• features (Union[str, List[Union[Tuple[str], str]], None], optional) – List of features to select, by default None

• dropna (bool, optional) – Whether to drop NaN values, by default False

Returns:

_description_

Return type:

pd.DataFrame

Wrapper for DataFrame which allows to manage metadata and perform validation

__init__(*args: Any, t: Optional[Union[DataFrame, TableDataset]] = None, **kwargs: Any) → None[source]#

Parameters:

t (optional) – pd.DataFrame or TableDataset to be set as table

__len__() → int[source]#

Returns length of the table

get(index: int) → Series[source]#

Returns a row from table by index

get_meta() → List[Dict[Any, Any]][source]#

Returns:

meta – A list where last element is this dataset’s metadata. Meta can be anything that is worth to document about the dataset and its data. This is done in form of list to enable cascade-like calls in Modifiers and Samplers.

Return type:

Meta

to_csv(path: str, **kwargs: Any) → None[source]#

Saves the table to .csv file. Any kwargs are sent to pd.DataFrame.to_csv.

Filter for table values

__init__(dataset: TableDataset, mask: List[bool], *args: Any, **kwargs: Any) → None[source]#

Parameters:

• dataset (TableDataset) – Dataset to be filtered.

• mask (Iterable[bool]) – Binary mask to select values from table.

Iterates over the table from path by the chunks.

__init__(csv_file_path: str, *args: Any, chunk_size: int = 1000, **kwargs: Any) → None[source]#

Parameters:

• csv_file_path (str) – Path to the .csv file

• chunk_size (int, optional) – number of rows to return in one __next__