audit-labs/audit-tools

A collection of scripts, queries, and other goodies you can use in an audit.

clone: git clone https://gitbay.org/audit-labs/audit-tools.git

440fcd1c889e8bc9075e9213fb19da14030ae327

verified · cmc

author: Christian Cleberg <hello@cleberg.net> · 2026-08-07T02:13:56Z

Make sampling/sample.py reproducible and point to the real tool

Seed the example draw and note that audit_sample.py / sampling_tool records
population hash, seed, method, and version for defensible sampling.

 sampling/sample.py | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/sampling/sample.py b/sampling/sample.py
index 7f74285..11566e2 100644
--- a/sampling/sample.py
+++ b/sampling/sample.py
@@ -1,5 +1,11 @@
 """
 Creates a sample from a CSV or Excel file based on user-defined SAMPLE_SIZE.
+
+NOTE: This is a minimal teaching snippet. For real fieldwork use the
+`audit_sample.py` CLI (or the `sampling_tool` package), which records the
+population hash, seed, method, and tool version in a manifest so the sample is
+reproducible and defensible. This file fixes a SEED only so the example itself is
+repeatable; it does not emit that provenance.
 """
 
 # Import packages
@@ -8,6 +14,10 @@ import pandas as pd
 # Define the sample size
 SAMPLE_SIZE = 25
 
+# A fixed seed makes the draw reproducible: same population + same seed => same
+# rows. Record the seed alongside any sample you rely on.
+SEED = 20260707
+
 # Import the data to a pandas DataFrame
 df = pd.read_csv("FILENAME_GOES_HERE.csv")
 
@@ -19,7 +29,7 @@ df = pd.read_csv("FILENAME_GOES_HERE.csv")
 print("Dataframe size (rows, columns): ", df.shape)
 
 # Sample
-sample = df.sample(SAMPLE_SIZE)
+sample = df.sample(SAMPLE_SIZE, random_state=SEED)
 print("Sample size: ", SAMPLE_SIZE)
 print("Sample:\n", sample)
 
@@ -31,4 +41,4 @@ print("Sample:\n", sample)
 #
 # # Sample Size: 25 + 5 replacement samples
 # SAMPLE_SIZE = 30
-# sample = df.sample(SAMPLE_SIZE, replace=True)
+# sample = df.sample(SAMPLE_SIZE, replace=True, random_state=SEED)