</>
pythonintermediate

Stratified Sampling with pandas

Draw a stratified random sample from a DataFrame, preserving class proportions for ML splits.

python
import pandas as pd

df = pd.DataFrame({'class':['A']*500 + ['B']*300 + ['C']*200, 'value': range(1000)})

def stratified_sample(df, col, n, seed=42):
    return (
        df.groupby(col, group_keys=False)
          .apply(lambda g: g.sample(frac=n/len(df), random_state=seed))
          .reset_index(drop=True)
    )

sample = stratified_sample(df, 'class', 100)
print(sample['class'].value_counts())

Use Cases

  • ML dataset splitting
  • balanced sampling
  • survey sampling

Tags

#pandas#sampling#stratified#ml

Related Snippets

Similar patterns you can reuse in the same workflow.

pythonbeginner

Pandas Categorical Encoding for ML

One-hot encode, label encode, and ordinal encode categorical columns using pandas and scikit-learn.

Best for: ML preprocessing

#pandas#encoding
pythonintermediate

Pandas Cartesian Feature Interaction

Generate pairwise feature interactions for ML by creating cross-product columns.

Best for: ML feature engineering

#pandas#feature-engineering
pythonbeginner

Pandas DataFrame Transformations

Common pandas DataFrame transformations including column operations, type casting, and string methods.

Best for: Cleaning raw data files for analysis

#pandas#dataframe
pythonbeginner

Pandas DataFrame Filtering Techniques

Filter DataFrames using boolean masks, query syntax, isin, between, and string matching methods.

Best for: Extracting subsets of data for reporting

#pandas#filtering