lab-data-utils 0.2.0

Experimental data utilities: error propagation and LaTeX table generator

lab-data-utils

Laboratory Data Utilities

A Python library for experimental data analysis and reporting.
Includes tools for outlier detection, error propagation and LaTeX-ready table formatting.

Features

• Error propagation via finite difference derivatives, with correct significant figures.
• LaTeX table generation with support for merged value ± error columns.
• Designed for clarity and reproducibility in lab reports.

Installation

pip install labdatautils

---

Function Reference

1. propagate(func, *args)

Arguments:

• func (callable) — Function whose result and uncertainty are to be calculated.
• *args — Values and uncertainties, alternating: val1, err1, val2, err2, ...
  Each can be a float, list, or NumPy array (all arrays must have the same length).

Returns:

• values (ndarray) — Computed function results.
• errors (ndarray) — Propagated uncertainties.

---

2. generate_latex_table(column_headers, data_columns, **kwargs)

Arguments:

• column_headers (list of str) — Ordered list of column names to display.
• data_columns (sequence of arrays) — One array per column, in the same order as column_headers.
  Each array should contain the values for that column.
• caption (str, optional) — Table caption. Default: "Data Table".
• label (str, optional) — LaTeX label for referencing. Default: "tab:data".
• error_map (dict or list of str, optional) — Maps error columns to their associated value column.
  Example: {"dx": "x", "dy": "y"} or ["dx:x", "dy:y"].
• merge_error_style (str, optional) — How to display value and error pairs:
  "separate" (two columns), "pm" (e.g. 1.23 ± 0.04), "paren" (e.g. 1.23(4)). Default: `"separate"

3. Outlier Detection

The library provides three layers of functionality for multiple methods, z-score, modified (MAD) z-score, IQR and Grubb's test:

---

a) compute_* (e.g. compute_zscores(x, **kwargs))

Arguments:

• x (array-like) — Input data.
• **kwargs — Method-specific options (e.g. ddof for z-scores, threshold for IQR).

Returns:

• scores (ndarray) — Raw scores for each point (z-scores, MAD-scores, IQR-scores, or Grubbs’ G values).

---

b) detect_outliers_* (e.g. detect_outliers_zscore(x, **kwargs))

Arguments:

• x (array-like) — Input data.
• threshold (float, optional) — Criterion for outliers. Default depends on method
  (3.0 for z, 3.5 for MAD, 1.5 for IQR, significance level α=0.05 for Grubbs).
• verbose (bool, optional) — If True, prints detected outliers. Default: True.

Returns:

• mask (ndarray of bool) — Boolean array, True where the input is an outlier.

---

c) remove_outliers_* (e.g. remove_outliers_iqr(x, **kwargs))

Arguments:

• x (array-like) — Input data.
• threshold (float, optional) — Same meaning as in detect_outliers_*.
• verbose (bool, optional) — If True, prints which points were removed. Default: True.

Returns:

• cleaned (ndarray) — Input data with outliers removed.

---

General entry points

Instead of calling method-specific functions, you can use the general function which have the different methods as arguments, for example:

from lab_data_utils import detect_outliers, remove_outliers

mask = detect_outliers(data, method="iqr", threshold=1.5)
cleaned = remove_outliers(data, method="iqr", threshold=1.5)