Sunday, December 17, 2017

Getting started with Python Pickle Module for saving objects (serialization)

pandas.DataFrame.to_pickle Pickle can serialize any python objects. Everything in Python is an object so technically many python components can be pickled. The concept is we can save any object and data objects in python into pickle format, of which we can load later, much like csv. This is a very important data science tool, and is especially useful for working with pandas in machine learning and deep learning.

Turn a variable called my_dataframe, my variable name for a Pandas Dataframe into a pickle file with extension pkl.

This tutorial introduces the idea of pickling of Python objections, Python Dictionaries for better loading speed, optimization and note pickling is used in Machine Learning as well.

Unlike other serialization libraries, pickle can serialize (flatten or "preserve") most python objects. It is also different from JSON, which is human readable. Pickle is binary, hence it is fast to load and small to store.

Check out the official documentation for this detail. Source 1

Source 1

No comments:

Post a Comment

React UI, UI UX, Reactstrap React Bootstrap

React UI MATERIAL  Install yarn add @material-ui/icons Reactstrap FORMS. Controlled Forms. Uncontrolled Forms.  Columns, grid