dataframes with numpy and pandas

Tables are where most applied ML work starts. Before a model trains, someone has to decide what counts as a row, which columns are allowed, how missing values behave, and whether the same transformation can run at inference time.

This chapter uses browser-safe Python lists and dictionaries to model the habits behind NumPy arrays and pandas DataFrames. The real packages matter, but the first engineering muscle is simpler: inspect shape, select columns deliberately, clean types, aggregate features, and prove the table is model-ready.

The mission thread ends with an API-to-dataframe pipeline: collect records, normalize them, reject bad rows, and return a small feature table that a later model could consume.