viper

Simple, expressive pipeline syntax to transform and manipulate data with ease

Overview

viper is a Python package that provides a simple, expressive way to work with data. It allows you to easily manipulate and transform data using a pipeline syntax similar to that of dplyr.

Pipelining your DataFrame manipulation operations offers several benefits:

• improved code readability (no need to 'comment the what')
• no need to save intermediate dataframes
• ability to chain a long sequence of operations in a single command
• thinking of coding as a series of transformations between the input and the desired output can improve the design and make it less coupled

Docs

Complete documentation and reference are available on the package's site.

Quick Start

Installation:

pip install viper

Here is an example of how to use viper to analyze the famed mtcars dataset.

We want to find:

• the average consumption, expressed in Miles/(US) gallon
• the average power

Furthermore:

• only consider those cars that weigh more than 2000lbs
• group the results by the number of cylinders and number of gears
• arrange in descending orders by the grouping variables

from viper.main import *
from viper.data import mtcars

pipeline(
    mtcars,
    rename(
        "hp = power",
        "mpg = consumption",
    ),
    mutate(
        consumption=lambda r: 1 / r["consumption"]
    ),
    filter(
        lambda r: r["wt"] > 2
    ),
    group_by("cyl", "gear"),
    summarize(
        "power = mean()",
        "consumption = mean()"
    ),
    arrange(
        "cyl desc",
        "gear desc"
    ),
)
#                power  consumption
# cyl gear
# 8   5     299.500000     0.064979
#     3     194.166667     0.068824
# 6   5     175.000000     0.050761
#     4     116.500000     0.050875
#     3     107.500000     0.050989
# 4   5      91.000000     0.038462
#     4      85.000000     0.041259
#     3      97.000000     0.046512

Here you can find more examples, particularly on joins.

Roadmap

The future development of the package will probably focus on:

• adding pivot_longerand pivot_wider functions
• adding more join_* functions

Contributions

You are welcome to contribute to the project or open issues if you have any ideas.