AddTukeyOutlierColumn — Analysis Tool Box

Introduction

Before trusting an average or feeding a column into a model, it helps to know whether a handful of extreme values are quietly distorting it. AddTukeyOutlierColumn answers "which values in this column are statistically unusual, using a method that doesn't assume a normal distribution?" by flagging values that fall outside Tukey's IQR fence — a robust, non-parametric boundary — and optionally visualizing the distribution with those boundaries marked. Reach for it for data quality checks, fraud or error detection, or any exploratory pass where you need a defensible, distribution-free definition of "unusual."

Teaching Note

The function is particularly useful for:

Data quality assessment and anomaly detection
Identifying extreme values in financial and scientific data
Preprocessing data for statistical modeling
Finding fraudulent transactions or unusual patterns
Quality control in manufacturing and operations
Exploratory data analysis and distribution understanding
Detecting measurement errors or data entry mistakes

Parameters

Parameter	Default	Description
`dataframe`required	—	A pandas DataFrame containing the data to analyze for outliers. A new column will be added to flag outlier values.
`value_column_name`required	—	Name of the numeric column to check for outliers.
`tukey_boundary_multiplier`	`1.5`	The multiplier for the IQR to determine outlier boundaries. Common values are 1.5 (standard outliers) or 3.0 (extreme outliers).
`plot_tukey_outliers`	`False`	If True, generates a histogram visualization showing the distribution with outliers highlighted in red and boundary lines marked.
`fill_color`	`'#999999'`	Hex color code for non-outlier histogram bars when plotting. Only used if plot_tukey_outliers=True.
`fill_transparency`	`0.6`	Alpha transparency value for histogram bars (0.0 to 1.0). Only used if plot_tukey_outliers=True.
`title_for_plot`	`'Tukey Outliers'`	Main title text for the plot. Only used if plot_tukey_outliers=True.
`subtitle_for_plot`	`'Shows the values that are Tukey outliers.'`	Subtitle text displayed below the title. Only used if plot_tukey_outliers=True.
`caption_for_plot`	`None`	Optional caption text displayed at the bottom of the plot. Only used if plot_tukey_outliers=True.
`data_source_for_plot`	`None`	Optional data source attribution text appended to the caption. Only used if plot_tukey_outliers=True.
`show_y_axis`	`False`	If True, displays the y-axis (frequency counts) on the plot. Only used if plot_tukey_outliers=True.
`title_y_indent`	`1.1`	Vertical position adjustment for the title as a fraction of the plot height. Only used if plot_tukey_outliers=True.
`subtitle_y_indent`	`1.05`	Vertical position adjustment for the subtitle as a fraction of the plot height. Only used if plot_tukey_outliers=True.
`caption_y_indent`	`-0.15`	Vertical position adjustment for the caption as a fraction of the plot height. Only used if plot_tukey_outliers=True.
`figure_size`	`(8, 6)`	Tuple specifying the (width, height) of the plot in inches. Only used if plot_tukey_outliers=True.

Returns

The original DataFrame with an additional binary column named {value_column_name} - Tukey outlier, where 1 indicates an outlier and 0 indicates a normal value. If plot_tukey_outliers=True, a visualization is also displayed.

Example

python

from analysistoolbox.data_processing import AddTukeyOutlierColumn
import pandas as pd

# Detect outliers in transaction amounts with default settings
transactions = pd.DataFrame({
    'transaction_id': range(1, 11),
    'amount': [100, 150, 120, 130, 140, 110, 125, 1500, 135, 145]
})
transactions = AddTukeyOutlierColumn(transactions, 'amount')
# Adds column 'amount - Tukey outlier' with 1 for the $1500 transaction