Mixed Data Cleaning & Analysis
Budget: $10 – $30 USD
I’m preparing a machine-learning and NLP pipeline and need a solid, reproducible cleaning and exploratory-analysis pass on a mixed dataset that combines structured tables with free-text fields. The work must be carried out in Python, using Pandas and NumPy as the core libraries; feel free to pull in complementary packages (e.g., scikit-learn, spaCy) where that will speed things up, but the final code has to run end-to-end in a clean environment that I can recreate with a single requirements file.
Here’s what I’m aiming for:
• A well-commented Jupyter notebook or .py script that ingests the raw data, handles missing values and outliers, normalises categorical variables, and applies common text-processing steps (tokenisation, stop-word removal, lower-casing, etc.).
• A set of intermediate and final CSV/parquet outputs representing each major stage so I can inspect the transformations.
• A concise summary (Markdown is fine) explaining every key decision, from regexes used on the text fields to any assumptions you made when merging tables.
I selected Python because it plugs directly into the rest of my stack, yet if you have a clever R or Excel trick that saves time, just flag it in your proposal and show how you’ll integrate it back into the Python workflow.
Submit a detailed project proposal that walks me through your planned approach, milestones and the checks you’ll use to validate data quality. Past work is helpful, but I’ll be choosing mainly on the clarity and realism of that proposal.
Here’s what I’m aiming for:
• A well-commented Jupyter notebook or .py script that ingests the raw data, handles missing values and outliers, normalises categorical variables, and applies common text-processing steps (tokenisation, stop-word removal, lower-casing, etc.).
• A set of intermediate and final CSV/parquet outputs representing each major stage so I can inspect the transformations.
• A concise summary (Markdown is fine) explaining every key decision, from regexes used on the text fields to any assumptions you made when merging tables.
I selected Python because it plugs directly into the rest of my stack, yet if you have a clever R or Excel trick that saves time, just flag it in your proposal and show how you’ll integrate it back into the Python workflow.
Submit a detailed project proposal that walks me through your planned approach, milestones and the checks you’ll use to validate data quality. Past work is helpful, but I’ll be choosing mainly on the clarity and realism of that proposal.