Pandas Data Analysis taste — ChatGPT prompt: what it does
The Pandas Data Analysis taste prompt for ChatGPT and Claude.ai offers a set of best practices for working with the Pandas library. It helps apply principles of reproducible data analysis.
CC0-1.0 ★ 178,201 updated 2026-10-24
Open on GitHubWhat Pandas Data Analysis taste does
The "Pandas Data Analysis taste" prompt offers a set of principles for creating reproducible data analysis using the Pandas library. It focuses on practices that help avoid common pitfalls and improve code reliability. Key recommendations include keeping analysis in version-controlled scripts, pinning library versions (pandas, numpy, pyarrow) in a lockfile for consistent results, and explicitly specifying data types (dtype, parse_dates, usecols) when reading large files, rather than letting Pandas guess them.
The prompt advises preferring the Parquet format for intermediate datasets due to its type preservation, good compression, and faster loading. It also recommends using pyarrow-backed string and nullable dtypes to prevent missing values from silently converting integer columns to float. For data transformations, the prompt suggests using method chains with assign, pipe, and query, avoiding long sequences of in-place mutation. It is crucial to never use inplace=True, instead reassigning the result to maintain clear data flow. For data selection, .loc and .iloc should be used explicitly, avoiding chained indexing that triggers SettingWithCopyWarning. When dealing with large data volumes, the prompt proposes switching to DuckDB or Polars once Pandas data no longer fits comfortably in memory.
What it helps with
- Ensuring reproducibility in data analysis results
- Efficiently handling large datasets in Pandas
- Preventing silent data type shifts
- Maintaining clear data flow in transformations
- Validating assumptions after major data processing steps
Example prompts for the agent
- Provide best practices for reproducible data analysis using Pandas.
- Suggest guidelines for Pandas DataFrame operations, including type handling and merging.
- How to avoid
SettingWithCopyWarningwhen selecting data in Pandas?
How to install Pandas Data Analysis taste
How to use
Paste the text at the start of a chat or into project instructions
Manually: put the files in place
.claude/commands/prompts.csv.mdQuestions about Pandas Data Analysis taste
How to install Pandas Data Analysis taste in Claude Code?
In Claude Code, the prompt is installed by copying the prompts.csv.md file into the .claude/commands/ directory.
What does the Pandas Data Analysis taste prompt do?
This prompt provides a collection of recommendations for working with the Pandas library, aimed at creating reproducible and efficient data analysis. It covers topics like version control, data format selection, type handling, and testing.
What tools does this prompt recommend?
The prompt suggests using pyarrow for string and nullable dtypes, Parquet for intermediate datasets, DuckDB or Polars for large data, and pytest for testing cleaning functions.
Source text
prompts.csv · CC0-1.0
# Pandas Data Analysis taste of a reproducible analytics practitioner - Keep analysis in version-controlled scripts or notebooks with outputs cleared before commit, and move reusable logic into an importable module. Confidence: 0.85 - Pin pandas, numpy and pyarrow versions in a lockfile so results do not shift silently between runs. Confidence: 0.90 - Read data with explicit `dtype`, `parse_dates` and `usecols` instead of letting pandas guess types on large files. Confidence: 0.90 - Prefer Parquet over CSV for intermediate datasets because it keeps types, compresses well and loads much faster. Confidence: 0.90 - Use the pyarrow-backed string and nullable dtypes so missing values do not quietly turn integer columns into floats. Confidence: 0.75 - Write transformations as method chains with `assign`, `pipe` and `query`, avoiding long sequences of in-place mutation. Confidence: 0.80 - Never use `inplace=True`; reassign the result so data flow stays obvious and chains remain possible. Confidence: 0.85 - Select with `.loc` and `.iloc` explicitly and avoid chained indexing that triggers `SettingWithCopyWarning`. Confidence: 0.90 - Vectorize with built-in column operations and `np.where` or `np.select`; reach for `apply` with Python functions only as a last resort. Confidence: 0.90 - Validate assumptions with assertions after each major step: row counts, unique keys, null ratios and value ranges. Confidence: 0.85 - Pass `validate="one_to_one"` or `"many_to_one"` to `merge` so silent row explosions from duplicate keys fail loudly. Confidence: 0.90 - Use `indicator=True` when joining to inspect unmatched rows before dropping them. Confidence: 0.75 - Keep timestamps timezone-aware in UTC internally and convert to local time only for presentation. Confidence: 0.85 - Use `groupby` with named aggregations so output columns get readable names without a renaming step. Confidence: 0.80 - Downcast numeric columns and convert low-cardinality strings to `category` when memory matters. Confidence: 0.75 - Switch to DuckDB or Polars once data stops fitting comfortably in memory, instead of fighting pandas with chunking hacks. Confidence: 0.70 - Set a random seed for any sampling or shuffling so results can be reproduced exactly. Confidence: 0.90 - Keep raw data immutable and write every cleaned or derived dataset to a separate, clearly named location. Confidence: 0.90 - Use snake_case column names, normalized once right after loading. Confidence: 0.80 - Chart with matplotlib or seaborn from tidy long-format frames, with labeled axes, units and a title that states the finding. Confidence: 0.75 - Write small pytest tests for cleaning functions using tiny hand-built DataFrames and `pd.testing.assert_frame_equal`. Confidence: 0.80
© f · CC0-1.0 · f/prompts.chat