https://www.oreilly.com/library/view/hands-on-exploratory-data/9781789537253/
Exploratory Data Analysis (EDA) is an approach to analyzing data sets to summarize their main characteristics, often with visual methods. Here are the steps involved in EDA:
-
Data Collection: This is the first step where you gather the data from various sources.
-
Data Cleaning: This step involves handling missing data, removing duplicates and outliers.
-
Data Wrangling or Data Munging: In this process, the raw data is transformed into another format to make it more suitable for analysis.
-
Data Analysis: In this step, various statistical techniques are used to analyze the data.
-
Data Visualization: This involves presenting your data in a visual format like graphs, charts etc., to understand the patterns, trends and insights in the data.
-
Summarize Data: Calculate basic statistics such as mean, median, and standard deviation to understand the central tendency and variability in the data.
-
Visualize Data: Create various plots and charts (histograms, scatter plots, box plots, etc.) to visualize patterns, trends, and outliers in the data.
-
Handle Missing Data: Investigate and address missing values in the dataset.
-
Identify Outliers: Detect and handle outliers that could skew the analysis.
-
Feature Engineering: Create new features or transformations of existing features that might be more useful for the analysis.
-
Data Cleaning: Address any data quality issues, such as duplicates or inconsistencies.
-
Hypothesis Testing: Formulate and test hypotheses to gain insights into the data.
import pandas as pd import numpy as np import matplotlib.pyplot as plt import seaborn as sns
df = pd.read_csv('data.csv')
print(df.head())
print(df.describe())
print(df.isnull().sum())
plot = sns.displot(df['ColumnName']) plt.show()
sns.boxplot(x=df['ColumnName']) plt.show()
Univariate Non-graphical: This is the simplest form of data analysis where we use just one variable to analyze the data. The standard goal of univariate non-graphical EDA is to understand the underlying sample distribution/data and make observations about the population1.
Multivariate Non-graphical: Multivariate non-graphical EDA technique is often used to show the relationship between two or more variables in the form of either cross-tabulation or statistics1.
Univariate Graphical: This involves visualizing a single variable’s data distribution using graphical methods like histograms, box plots, etc1.
Multivariate Graphical: This involves visualizing relationships between multiple variables using graphical methods like scatter plots, pair plots, etc1.
Effectivitiy depends on the specific scenario and what you’re trying to achieve with your analysis. Each type has its own strengths and is suited to different kinds of data and different questions. For example, if you’re trying to understand the distribution of a single variable, univariate graphical or non-graphical might be most effective. If you’re trying to understand relationships between variables, multivariate graphical or non-graphical might be more appropriate.
It’s also worth noting that these types are not mutually exclusive and can be used in combination during an EDA process. The key is to choose the right approach based on your data and your analysis goals.
Skweness and kurtosis - https://www.analyticsvidhya.com/blog/2021/05/shape-of-data-skewness-and-kurtosis/
EDA Tutorial From PYCON
https://github.com/cmawer/pycon-2017-eda-tutorial/tree/master