Skip to content

Examples

Practical examples in Jupyter Notebooks to learn DataFrameIt.

Available Notebooks

Examples are organized from basic to advanced. We recommend following in order.

1. Basic

01_basic.ipynb

Introduction to DataFrameIt with sentiment analysis.

  • Create simple Pydantic model
  • Process basic DataFrame
  • Understand the output
from pydantic import BaseModel
from typing import Literal

class Sentiment(BaseModel):
    sentiment: Literal['positive', 'negative', 'neutral']

2. Error Handling

02_error_handling.ipynb

How to handle errors and configure retry.

  • Configure max_retries, base_delay, max_delay
  • Check _dataframeit_status column, which only appears when some row fails
  • Analyze _error_details

3. Incremental Processing

03_resume.ipynb

Continue processing from where it stopped.

  • Save checkpoints with batch_size and checkpoint_path
  • Reload with read_df and continue with resume=True
  • Reprocess only error rows

4. Prompt and Text Column

04_custom_placeholder.ipynb

Control where text appears in the prompt and which column it comes from.

  • Place {texto} in the template, or let the text go at the end
  • Inference of text_column and when to set it
  • Rows with empty text

05_advanced_legal.ipynb

Real example with complex Pydantic model.

  • Nested models
  • Optional fields
  • Lists of objects
  • Multiple entity extraction
class Party(BaseModel):
    name: str
    type: Literal['plaintiff', 'defendant']

class Decision(BaseModel):
    parties: List[Party]
    outcome: Literal['granted', 'denied']

6. Polars

06_polars.ipynb

Use DataFrameIt with Polars instead of Pandas.

  • Input with polars.DataFrame
  • Output preserves Polars type

7. Multiple Data Types

07_multiple_data_types.ipynb

Process different input types.

  • Lists
  • Dictionaries
  • Series

8. Rate Limiting

08_rate_limiting.ipynb

Control request rate.

  • Configure rate_limit_delay
  • Use parallel_requests
  • Combine for maximum efficiency

example_09_web_search.py

Script with agent-based web search.

  • use_search, search_per_field and save_trace
  • Tavily or Exa as the search provider

Running the Examples

1. Clone the Repository

git clone https://github.com/bdcdo/dataframeit.git
cd dataframeit

2. Install Dependencies

pip install dataframeit[openai]
pip install jupyter

3. Configure your API Key

export OPENAI_API_KEY="your-key"

4. Run Jupyter

jupyter notebook example/

Contributing Examples

If you created an interesting example, consider contributing! Open an issue or pull request on GitHub.