What’s missing?
Find empty cells in every column. See the counts and percentages before deciding how to handle them.
Understand completeness ↗ A CLEARER START FOR YOUR DATA
Confidence starts
at the baseline.
Before the model. Before the conclusion.
Know what’s actually in your data.
A lightweight, open-source toolkit that turns a CSV into a transparent, reproducible data quality report.
v0.1.0· Python 3.9+ · No dependencies
+ INSIDE THE EXAMPLE DATASETFIG. 01
Every observation tells a story.
Start by checking the whole picture.
sample_id
100.0%
time_s
100.0%
temperature_c
99.2%
pressure_kpa
99.2%
signal
98.8%
quality
100.0%
Observed signalSynthetic example · not research data
OBSERVATIONS240rows
MISSING CELLS7to review
DUPLICATE ROWS2to review
observations.csvExplore the report ↗
Your data stays on your machine
Source you can inspect
Results you can trace
MIT licensed
01 / THE APPROACHLESS ASSUMPTION. MORE INSPECTION.
Good analysis begins
with better questions.
Small data issues can quietly shape a conclusion. Baseline makes the first checks explicit, so you can make the next decision with context.
What’s repeated?
Identify exact duplicate rows. Count repeated records before deciding whether they belong.
Understand duplication ↗What’s the baseline?
Summarize numeric columns and record the file’s SHA-256 fingerprint. Keep the result tied to its input.
Understand the output ↗02 / A LOOK INSIDEREAL OUTPUT. SYNTHETIC DATA.
Nothing hidden.
Everything inspectable.
A useful report doesn’t ask for your trust. It shows you what was checked, what was found, and which file it came from.
↳
View the complete JSON report ↗This example has 7 missing cells and 2 extra duplicate rows. The report records them for your review; your original file stays untouched.
BASELINE REPORTobservations.csv
A few things worth a closer look.
Inspection complete. Review the findings below.
Dataset dimensions240 rows × 6 columnsRecorded
Empty cells7 / 1,440 cellsReview
Exact duplicate rows2 extra rowsReview
Input fingerprintSHA-256Recorded
Empty or whitespace-only cells, by column.
| Column | Missing | Rate |
|---|---|---|
| sample_id | 0 | 0.00% |
| time_s | 0 | 0.00% |
| temperature_c | 2 | 0.83% |
| pressure_kpa | 2 | 0.83% |
| signal | 3 | 1.25% |
| quality | 0 | 0.00% |
An exact fingerprint of the bytes that were inspected.
- Tool
- Baseline v0.1.0
- Input
- observations.csv
- Algorithm
- SHA-256
INPUT FILE HASH
bb0e2fb30eb5272f4bb7acba12710b7505d3cc1563b4349f154f0afc7d690854A matching hash identifies the same input bytes. It does not certify the data’s accuracy.
JSON output · Portable by designOutput reference ↗
03 / FROM FILE TO FINDINGSONE SCRIPT. NO EXTRA PACKAGES.
Your first report.
Three simple steps.
Start with the included example.
Then bring your own CSV.
- 01
Download & unpack
The script, example data, and a readable guide.
- 02
Read & run
Inspect the source. Run it with Python 3.9+.
- 03
Review your report
Open the JSON output and follow the findings.
QUICKSTART / macOS & LINUX
# 1. In the folder containing your download
unzip baseline-starter-v0.1.0.zip
cd baseline-starter# 2. Run the included example
python3 baseline.py \
examples/observations.csv \
--output report.json✓ Report written to report.json
# Next: open report.json in your text editor.
Your input file is never modified.
Python standard library onlyWindows instructions ↗
A SMALL STEP TOWARD BETTER EVIDENCE
Start with what you can verify.
Download the source. Follow the method. Know your baseline.