Pandas — анализ и обработка данных

data

Pandas pro is a Claude Code skill that handles pandas DataFrame operations for data analysis, cleaning, transformation, aggregation, merging, time series processing, and memory optimization on large datasets. Reach for it when you need to join tables on composite keys without losing rows, when raw data is messy and needs null imputation or deduplication before analysis, when groupby or pivot_table pipelines are slow and need vectorization, or when timestamped data must be resampled to daily or weekly intervals. Inside are production-grade patterns: vectorized assignments instead of iterrows, safe subset mutation with .copy(), merge with key validation and indicator columns, forward-fill and linear interpolation for missing values, time series resampling, and memory reduction via categorical types and numeric downcast. The skill outputs Python code with inline validation checks on dtypes, shapes, and null counts after every transformation. The direction is in-memory tabular data manipulation, not querying databases or distributed processing — for broader Python tasks use the related python-pro skill. Works for any data wrangling complexity as long as the dataset fits in RAM; not suitable when data exceeds memory and Dask or Spark is needed.