verticapy.vDataFrame.outliers#

vDataFrame.outliers(columns: str | list[str] | None = None, name: str = 'distribution_outliers', threshold: float = 3.0, robust: bool = False) → vDataFrame#

Adds a new vDataColumn labeled with 0 or 1, where 1 indicates that the record is a global outlier.

Parameters#

columns: SQLColumns, optional: List of the vDataColumn names. If empty, all numerical vDataColumn are used.
name: str, optional: Name of the new vDataColumn.
threshold: float, optional: Threshold equal to the critical score.
robust: bool: If set to True, uses the Robust Z-Score instead of the Z-Score.

Returns#

vDataFrame: self

Examples#

Let’s begin by importing VerticaPy.

import verticapy as vp

Hint

By assigning an alias to verticapy, we mitigate the risk of code collisions with other libraries. This precaution is necessary because verticapy uses commonly known function names like “average” and “median”, which can potentially lead to naming conflicts. The use of an alias ensures that the functions from verticapy are used as intended without interfering with functions from other libraries.

Let us create a vDataFrame that has some outliers:

import numpy as np

data = np.random.normal(
    loc = 0,
    scale = 1,
    size = 10,
)


data = np.append(data, [100])

vdf = vp.vDataFrame({"vals": data})

	123 vals Numeric(22) 100%
1	-0.9692907406301432
2	1.5361282477773475
3	-0.845487667550313
4	0.8760238645325606
5	-2.1440816057858076
6	0.6047497570245369
7	0.11152711520356214
8	-0.9210417560931947
9	0.8408167674806595
10	1.1481774819371722
11	100.0

Now we can see which values are outliers by using the vDataFrame.outliers() method:

vdf.outliers()

	123 vals Numeric(22) 100%	123 distribution_outliers Integer 100%
1	-0.9692907406301432	0
2	1.5361282477773475	0
3	-0.845487667550313	0
4	0.8760238645325606	0
5	-2.1440816057858076	0
6	0.6047497570245369	0
7	0.11152711520356214	0
8	-0.9210417560931947	0
9	0.8408167674806595	0
10	1.1481774819371722	0
11	100.0	1

Note

This function can only identify global outliers in the distribution. For other types of outliers, it is recommended to create machine learning models.