framex 0.4.0

dataframe operations

FrameX

Framex is a light-weight, dataset fetching library for fast prototyping, tutorial creation, and experimenting.

Built on top of Polars.

Contents

• Installation
• Usage
  • Python
    • Loading datasets
    • Listing available datasets
    • Getting information on Datasets
    • Getting Dataset URLs
  • CLI

Installation

To get started, install the library with:

pip install framex

Usage

Python

import framex as fx

Loading datasets

iris = fx.load("iris")

which returns a polars DataFrame
Therefore, you can use all the polars functions and methods on the returned DataFrame.

iris.head()

shape: (5, 5)
┌──────────────┬─────────────┬──────────────┬─────────────┬─────────┐
│ sepal_length ┆ sepal_width ┆ petal_length ┆ petal_width ┆ species │
│ ---          ┆ ---         ┆ ---          ┆ ---         ┆ ---     │
│ f32          ┆ f32         ┆ f32          ┆ f32         ┆ str     │
╞══════════════╪═════════════╪══════════════╪═════════════╪═════════╡
│ 5.1          ┆ 3.5         ┆ 1.4          ┆ 0.2         ┆ setosa  │
│ 4.9          ┆ 3.0         ┆ 1.4          ┆ 0.2         ┆ setosa  │
│ 4.7          ┆ 3.2         ┆ 1.3          ┆ 0.2         ┆ setosa  │
│ 4.6          ┆ 3.1         ┆ 1.5          ┆ 0.2         ┆ setosa  │
│ 5.0          ┆ 3.6         ┆ 1.4          ┆ 0.2         ┆ setosa  │
└──────────────┴─────────────┴──────────────┴─────────────┴─────────┘

iris = fx.load("iris", lazy=True)

which returns a polars LazyFrame

Both these operations create local copies of the datasets by default cache=True.

Available datasets

To see the list of available datasets, run:

fx.available()

{'remote': ['iris', 'mpg', 'netflix', 'starbucks', 'titanic'], 'local': ['titanic']}

which returns a dictionary of both locally and remotely available datasets.

To see only local or remote datasets, run:

fx.available("local")
fx.available("remote")

{'local': ['titanic']}
{'remote': ['iris', 'mpg', 'netflix', 'starbucks', 'titanic']}

Getting information on Datasets

To get information on a dataset, run:

fx.about("mpg") # basically the same as `fx.about("mpg", mode="print")`

which will print the information on the dataset as the following:

NAME    : mpg
SOURCE  : https://www.kaggle.com/datasets/uciml/autompg-dataset
LICENSE : CC0: Public Domain
ORIGIN  : Kaggle
OG NAME : autompg-dataset

Or you can get the information as a single row polars.DataFrame by running:

row = fx.about("mpg", mode="row")
print(row)

which will print the information on the dataset ASCII art as the following:

shape: (1, 4)
┌──────┬─────────────────────────────────┬────────────────────┬────────┐       
│ name ┆ source                          ┆ license            ┆ origin │       
│ ---  ┆ ---                             ┆ ---                ┆ ---    │       
│ str  ┆ str                             ┆ str                ┆ str    │       
╞══════╪═════════════════════════════════╪════════════════════╪════════╡       
│ mpg  ┆ https://www.kaggle.com/dataset… ┆ CC0: Public Domain ┆ Kaggle │       
└──────┴─────────────────────────────────┴────────────────────┴────────┘ 

or you can simply treat row as a polars DataFrame in your code.

Getting Dataset URLs

In case you need the file links.

url_pokemon = fx.get_url("pokemon")

by default, the format is " feather".

Optionally, you can specify the format of the dataset.

url_pokemon_csv = fx.get_url("pokemon", format="csv")

CLI

Get a single dataset:

fx get iris

or get multiple datasets:

fx get iris mpg titanic

which will download dataset(s) to the current directory.

For more parameters

fx get --help

To get the name of the available datasets on the remote server.

fx list

this will list all available datasets on the remote server.