Python data-cleaning and wrangling exercises maintained by Leela Padala. The notebooks demonstrate data inspection, missing-value handling, numeric normalization, grouped summaries, and bootstrap analysis with pandas and NumPy.
| Notebook | Implemented work | Required data |
|---|---|---|
| Data Cleaning | Missing-value counts, dropping heavily incomplete columns, median/mode imputation | train.csv (not bundled) |
| Data Analysis | Unit stripping, numeric conversion, binary category mapping, vehicle-age calculation, filtering and grouped summaries | train.csv (not bundled) |
| Data Wrangling Assignment | Schema inspection, selection, filtering, aggregation, and loan/housing data summaries | External CSV URLs embedded in the notebook |
| Sampling and Bootstrap Exercise | A fixed-seed sample comparison, BMI percentile, and repeated bootstrap summaries for blood pressure | diabetes.csv (not bundled) |
Open the relevant notebook in Google Colab or a Jupyter environment with pandas, NumPy, matplotlib, and seaborn. Put the required CSV in the working directory before running its cells. For the wrangling assignment, allow network access to its referenced public datasets.
The vehicle and diabetes exercises are not self-contained because their original input files are absent. Do not substitute an arbitrary dataset: its schema and meaning may differ. Notebook code and historical outputs are preserved.
- Inspect types, missing values, and category values before and after transformations.
- The vehicle notebook implements binary mappings, not a full one-hot encoder; unmapped categories become missing values.
- Vehicle age depends on the current year, and price-per-kilometer calculations need review for zero distances.
- Median/mode imputation needs additional handling for entirely empty columns.
- External URLs are not version-pinned and may change.
- Bootstrap results are educational summaries, not medical conclusions or a validated population model.
The portfolio documentation review validated notebook JSON, referenced filenames, and data-source links. No fresh execution or automated test coverage is claimed. These projects demonstrate supporting Python/data skills rather than production cybersecurity capabilities.
The wrangling notebook references LoanPredictionProblem and the CSV.jl Sacramento real-estate test dataset. Existing source references are retained. Provenance for the missing vehicle and diabetes CSVs remains unverified.