DataFrames: The Universal Table

A DataFrame is a table: rows are examples, named columns are features. Build one from a dict of lists:

• df = pd.DataFrame({'model': ['a', 'b'], 'acc': [0.9, 0.7]})

Then run the universal first-contact ritual:

• df.head() — the first 5 rows
• df.shape — (rows, columns)
• df.info() — dtypes and non-null counts per column
• df.describe() — count, mean, std, min, quartiles, max for every numeric column

Ninety percent of exploratory analysis is those four calls plus curiosity. df.info() alone catches half of all data problems — a numeric column reading object means something inside isn't a number.