From 6226695a2072c2e269ece38254389e871c26e01f Mon Sep 17 00:00:00 2001
From: Christian Cleberg <hello@cleberg.net>
Date: Thu, 29 May 2025 11:41:39 -0500
Subject: feat: add stratified sampling script (#12)

* feat: add stratified sampling script

* Commit from GitHub Actions (Ruff)

---------

Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
---
 .gitignore                    |  3 ++-
 sampling/stratified_sample.py | 62 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 64 insertions(+), 1 deletion(-)
 create mode 100644 sampling/stratified_sample.py

diff --git a/.gitignore b/.gitignore
index 07ed393..840093e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,3 @@
-venv/
+.venv
+venv
 readme.html
diff --git a/sampling/stratified_sample.py b/sampling/stratified_sample.py
new file mode 100644
index 0000000..1ca94f9
--- /dev/null
+++ b/sampling/stratified_sample.py
@@ -0,0 +1,62 @@
+# Import packages
+import pandas as pd
+import math
+
+# Load data
+df = pd.read_csv("FILENAME_GOES_HERE.csv")
+
+# ALTERNATIVE: If you use Excel, use this instead. Supports xls, xlsx, xlsm,
+# xlsb, odf, ods and odt file extensions.
+# df = pd.read_excel("FILENAME_GOES_HERE.xlsx")
+
+# Print totals prior to sampling
+print("Dataframe size (rows, columns):", df.shape)
+
+# User-defined parameters
+SAMPLE_SIZE = 25
+STRATIFY_COLUMN = "Category"  # <- Change this to your column name
+
+# Define stratum proportions (as fractions)
+# Example: if you have categories A, B, and C
+stratum_proportions = {"A": 0.4, "B": 0.4, "C": 0.2}
+
+# Validate proportions sum to 1
+if not math.isclose(sum(stratum_proportions.values()), 1.0):
+    raise ValueError("Stratum proportions must sum to 1.")
+
+# Check that all strata exist in the data
+missing_strata = set(stratum_proportions.keys()) - set(df[STRATIFY_COLUMN].unique())
+if missing_strata:
+    raise ValueError(
+        f"Strata {missing_strata} not found in column '{STRATIFY_COLUMN}'."
+    )
+
+# Perform stratified sampling
+samples = []
+for stratum, proportion in stratum_proportions.items():
+    stratum_df = df[df[STRATIFY_COLUMN] == stratum]
+    n_samples = math.floor(SAMPLE_SIZE * proportion)
+    if n_samples > len(stratum_df):
+        raise ValueError(
+            f"Not enough data in stratum '{stratum}' to sample {n_samples} rows."
+        )
+    stratum_sample = stratum_df.sample(n=n_samples, random_state=42)
+    samples.append(stratum_sample)
+
+# Combine all stratum samples into one DataFrame
+final_sample = pd.concat(samples).reset_index()
+
+# If needed, randomly sample extra rows to fill any rounding gap
+current_sample_size = len(final_sample)
+if current_sample_size < SAMPLE_SIZE:
+    remaining = SAMPLE_SIZE - current_sample_size
+    remaining_sample = df.sample(n=remaining, random_state=42)
+    final_sample = pd.concat([final_sample, remaining_sample])
+
+# Print sample results
+print("Final sample size:", final_sample.shape[0])
+print("Sample breakdown by stratum:\n", final_sample[STRATIFY_COLUMN].value_counts())
+print("\nSample:\n", final_sample)
+
+# Optionally, save the sample to a new CSV
+# final_sample.to_csv("sample_output.csv", index=False)
-- 
cgit v1.2.3-70-g09d2