dtables / dtables

Python based minimalistic framework for exploratory data analysis.

dtables

dtables is a Python based minimalistic framework for exploratory data analysis. This project is under development. A rough API map is available.

Goals

• Minimal - Quick to learn and get going. Useful for programmers who are just dabbling in data analysis. Not necessarily for pros whose day job is data analysis. Focus on "one way to solve problems" rather than "different ways to match different situations".
• Complete - Minimal does not mean incomplete. Do not leave important use cases unaddressed. Try to address them without adding concepts.

Performance is not an initial goal. The current objective is to make it work well for smallish datasets (10s of MBs).

Features

Loaders

• load_dict
• load_tuples
• load_csv
• load_json
• load_table

Overview

• column names
• head
• tail
• dtypes
• shape
• describe

Column name Slicing

• name based
• position based

Dtable Slicing

Column Based

• single column name
• array of column names
• column name range
• single position
• array of position
• range of position
• boolean array based

Row based

• single position
• array of position
• range of position
• boolean array based

Assignment

• overwrite existing columns
• add new columns
• overwrite existing rows

Vectorized operations

• arithmetic
• comparison
• broadcasting

Transformation

• row apply
• column apply

Reshaping

• pivot
• melt

Extending

• append
• join

Aggregates

• sort
• group apply