Python Learning Journey

I am Devam Shah, learning Python, C++ DSA, and classical machine learning. This repository records small projects and experiments as I develop those skills.

Project 1: CSV Data Profiler

A standard-library command-line tool for inspecting a dataset before analysis. It reports row counts, missing cells, distinct nonmissing values, and numeric minimum, maximum, and mean. It rejects duplicate/blank headers and ragged rows.

python csv_profiler.py examples/study_sessions.csv
python csv_profiler.py examples/study_sessions.csv --output report.json
python -m unittest discover -s tests -v

Requires Python 3.12; no third-party dependencies. See the synthetic sample data and its generated report.

Data rules and limitations

• Leading/trailing whitespace is trimmed; blank cells are missing.
• A numeric column must have only finite numeric nonmissing cells. Mixed text, NaN, and infinity cause the column to be treated as text.
• Distinct counts use trimmed text, so 1 and 1.0 are distinct cell strings.
• Empty files, duplicate headers, and inconsistent row widths give clear errors.
• Numeric cells and distinct values are retained in memory. This is a tool for small datasets, not streaming analysis of files larger than available RAM.
• Missingness does not automatically justify dropping rows or imputing values; those decisions depend on the dataset and analysis goal.

What to learn from this project

1. Trace how csv.reader handles quoted commas and blank lines.
2. Explain the difference between missing, empty, and nonfinite numeric values.
3. Run the tests, deliberately introduce a bug, and explain which test detects it.
4. Add a feature yourself, such as a configurable missing-value marker, with tests.

Related learning projects

• ML Fundamentals with NumPy
• Handwritten Digits Benchmark

Learning goals

• Improve Python programming and data handling.
• Practise Git branches, commits, pull requests, and code review.
• Understand classical ML algorithms and evaluate them correctly.
• Make useful, tested open-source contributions.

The initial CSV tool, tests, and documentation were prepared with Codex assistance. Future entries should record my own experiments and explanations. Repository activity is a learning record, not a claim of independent mastery.