atlantis 2021.9.13.1

Python library for simplifying data science

Atlantis

Atlantis is a Python library for simplifying programming with Python for data science.

Installation

You can just use pip to install Atlantis:

pip install atlantis

Modules

• collections helps with working with collections.
• colour simplifies using colours.
• ds (datascience) provides tools for:
  • data wrangling,
  • validation,
  • tuning,
  • sampling,
  • evaluation,
  • clustering, and
  • parallel processing of machine learning models.
• functions manages higher order functions.
• hash simplifies and standardizes hashing.
• text makes working with texts and strings easy.
• time
  • provides methods for interacting with time and date as well as
  • progress bars

collections

This module of the package atlantis helps with working with collections.

flatten

from atlantis.collections import flatten
flatten([1, 2, [3, 4, [5, 6], 7], 8])

returns: [1, 2, 3, 4, 5, 6, 7, 8]

List

This class inherits from Python's list class but implements a few additional functionalities.

from atlantis.collections import List
l = List(1, 2, 3, 4, 2, [1, 2], [1, 2])

Flattening:

l.flatten()
>>> List: [1, 2, 3, 4, 2, 1, 2, 1, 2]

Finding duplicates:

l.get_duplicates()
>>> List: [2, List: [1, 2]]

Note: the list elements of a List automatically get converted to Lists, recursively.

ds (Data Science)

This module provides data science tools for:

• data wrangling,
• validation,
• tuning,
• sampling,
• evaluation,
• clustering, and
• parallel processing of machine learning models.

KMeans Clustering

I have used the KMeans class from both sklearn and that of pyspark and was frustrated by two problems: (a) even though the two classes do exactly the same thing their interfaces are vastly different and (b) some of the simplest operations are very hard to do with both classes. I solved this problem by creating my own KMeans class that is a wrapper aroung both of those classes and uses the appropriate one automatically without complicating it for the data scientist programmer.

Clustering Optimization