If you are going to find out anything about a dataset,
then you must understand the data first and try to gather as many insights from it

What’s an EDA ?

Exploratory data analysis (EDA) is used by data scientists to analyze and investigate data sets and summarize their main characteristics, often employing data visualization methods.

Importance OF EDA

  • It helps in looking up data before making any assumptions.
  • It can help identify obvious errors, as well as better understand existing patterns within the data.
  • It helps in detecting outliers or anomalous events, find interesting relationships among the variables / features.
  • Deliver data-driven insights to business stakeholders.

Data & Its Types

The term “data” was first used to denote “transmissible and storable computer information” in 1946 (source).

Broadly Categorized Under Three Umbrella

  • Structured - ( For Example : Customer Information stored in RDBMS.. )
  • Semi-Structured - ( For Example : Json & XML Document, Emails, EDI.. )
  • Unstructured - ( For Example : Images, Audio Files, Video Files.. )

In this post we will focus on structured data, where I would walkthrough a systemic approach to quickly show latent statistics from one of the dataset.

Structured Data Types can further be categorized into 2 types :

1.) Categorical Data : Any data that is not a number. The rules of arithmetic do not apply.

Nominal - It comes from the latin word “nomen” which means Name.
It represents Simple Naming System.
For Example - Country Names
Ordinal - Data in which their order matters.
For Example Competitive Ranking 1,2,3..
Happiness Rating On Scale Of 10
Binary - Data which can have only 2 values.
For Example - Yes or No

2.) Quantitative Data : Data which can be represented in form of numbers. The rule of arithmetic operations can be applied on them.

Discrete - Numbers which have a logical end to them.
For Example - Interval, Days in a Month.
Continuous - Numbers which don’t have a logical end to them.
For Example - Heights.

Processes Involved In EDA

  • Data Collection - Gather data from different source

    Public Data Websites

    Social Websites

    Blogs / Websites etc via scrapping

    • JSoup
    • Beautiful Soup
  • Data Cleaning - Cleanse the data such as missing data or measurement error.

    Open Refine - https://openrefine.org/

  • Data Pre-processing

    Classic Unix Tools

    • sed/awk
    • Shell Scripts
    • GNU Parallel
  • Data Visualization - The objective of this technique is to display the important information in a clean, concise manner on a single screen with the purpose to inform and not misguide its readers.

    Some Tools For Data Visualization & Dashboard

    • Tableau
    • Apache Superset

Types Of EDA

  • Univariate Analysis
    This is the simplest form of data analysis, where the data being analyzed consists of only one variable.The main purpose of this analysis is to describe data and find patterns within it.
    Few Visualizations to perform Univariate Analysis :
    • Box Plot
    • Histogram
    • Stem and Leaf Plot
  • BiVariate Analysis
    This is similar to univariate and this analysis can be descriptive or inferential. It involves the analysis of the relationship between two variables.
    Few Visualizations to perform BiVariate Analysis :
    • Scatter Plot
    • Bivariate Line Chart
  • Multivariate Analysis
    Multivariate data analysis refers to any statistical technique used to analyze data that arises from more than one variable.
    Few Visualizations to perform Multivariate Analysis :
    • Scatter Plot
    • Bar Chart
    • Bubble Chart
    • Run Chart
  • Correlation Analysis

*Let’s Jump to Code Now *