vDataFrame.balance¶

In [ ]:

vDataFrame.balance(column: str, 
                   method: str = "hybrid", 
                   x: float = 0.5, 
                   order_by: list = [],)

Balances the dataset using the input method.

⚠ Warning: If the data is not sorted, the generated SQL code may differ between attempts.

Parameters¶

Name	Type	Optional	Description
column	str	❌	Column used to compute the different categories.
method	str	✓	The method with which to sample the data. hybrid : hybrid sampling. over : oversampling. under : undersampling.
x	float	✓	The desired ratio between the majority class and minority classes. Only used when method is 'over' or 'under'.
order_by	list	✓	vColumns used to sort the data.

Returns¶

vDataFrame : sample vDataFrame

Example¶

In [5]:

from verticapy.datasets import load_titanic
titanic = load_titanic()

# minority class is Q
titanic["embarked"].topk()

Out[5]:

	count	percent
S	873	70.746
C	253	20.502
Q	106	8.59

Rows: 1-3 | Columns: 3

In [7]:

# hybrid

balance = titanic.balance(column = "embarked")
balance["embarked"].topk()

Out[7]:

	count	percent
C	111	34.472
Q	106	32.919
S	105	32.609

Rows: 1-3 | Columns: 3

In [8]:

# over

balance = titanic.balance(column = "embarked", method = "over", x = 0.5)
balance["embarked"].topk()

Out[8]:

	count	percent
Q	106	47.964
S	61	27.602
C	54	24.434

Rows: 1-3 | Columns: 3

In [9]:

# under

balance = titanic.balance(column = "embarked", method = "under", x = 0.5)
balance["embarked"].topk()

Out[9]:

	count	percent
S	220	41.431
C	205	38.606
Q	106	19.962

Rows: 1-3 | Columns: 3