Scribd
Upload
10 views

Eda ML 2

Uploaded by

shailaja.m
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
10 views

Eda ML 2

Uploaded by

shailaja.m
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 10

Exploratory Data

Analysis(EDA)
Exploratory Data Analysis (EDA) is a critical process for understanding and gaining insights from your data. It involves a systematic approach to
examining and visualizing the key characteristics of a dataset, helping you uncover hidden patterns, identify anomalies, and drive informed decision-
making.

At the core of EDA is the need to become intimately familiar with your data. This involves reviewing the initial details about the data, such as data types,
missing values, and data distributions. From there, you can begin modifying or removing any unwanted data elements, ensuring your analysis is focused on
the most relevant information.

With a clean dataset in hand, the next step is to retrieve and organize the data in a way that enables effortless exploration. This may involve aggregating,
filtering, or transforming the data to uncover meaningful insights. Generating statistical measures, such as averages, correlations, and standard deviations,
can further illuminate the underlying patterns and relationships within the data.

Finally, the power of EDA lies in its ability to visually represent the data through graphs, plots, and other visualizations. These graphical representations
can help you identify trends, outliers, and opportunities for deeper investigation, ultimately leading to a more comprehensive understanding of your data.
The Process of EDA
Exploratory Data Analysis (EDA) is an essential approach for analyzing
datasets and uncovering their key characteristics, often using visual methods.
EDA is crucial in the early stages of a machine learning project, as it helps
you understand the data you'll be working with, identify patterns, spot
anomalies, and check assumptions before moving on to modeling.
Knowing Initial Details about
Data
1 Understanding the Data
Get familiar with your dataset, including the information it contains and what each feature
(column) represents. This includes understanding the shape of the data, data types, and
column names.

2 Exploring Data Types


Determine whether features are numerical (like age or salary) or categorical (like job title or
gender), as this will inform your analysis and modeling approaches.

3 Identifying Columns
Understand what each column represents, as this gives you insight into the kind of data
you're working with and how it might be relevant to your analysis.
Modifying or Removing Unwanted Data
Handling Missing Values Removing Duplicates Filtering Outliers
Some data points might be missing. If the same data point is recorded more Outliers are extreme values that can
You can choose to fill them in than once, you'll need to clean up the skew your analysis. Deciding whether
(imputation) or remove those dataset to avoid biases. to keep or remove them depends on
rows/columns, depending on your their relevance to your analysis.
analysis needs.
Retrieving Data
Selecting Specific Features Accessing Data
Focus on the columns relevant to your analysis, such as Retrieve the relevant information from your dataset, ensuring
'Salary' and 'Years of Experience'. you have the data you need for your analysis.

1 2 3

Subsetting Data
Create smaller datasets that include only specific rows that
meet certain conditions, like employees from a specific
department.
Getting Statistical Data
Descriptive Statistics
Calculate measures like mean, median, mode, standard deviation, and range to
summarize the dataset and understand its distribution.

Distribution Analysis
Analyze how values are distributed (e.g., normal distribution, skewed
distribution) to inform your modeling choices.

Identifying Patterns
Uncover trends and patterns in the data that can provide valuable insights for
your analysis and modeling.
Drawing Graphs/Plots

Histograms Box Plots Scatter Plots Bar Charts


Visualize the distribution of Useful for visualizing the Help in identifying Ideal for comparing
numerical data, showing spread of the data and relationships between two categorical data, showing the
how often different ranges of identifying outliers. numerical variables (e.g., frequency of each category.
values occur. years of experience vs.
salary).
The Importance of EDA

Understanding Data
EDA helps you thoroughly understand the data you're working with, which is crucial for making informed decisions
in your analysis and modeling.

Identifying Patterns
EDA allows you to uncover patterns, trends, and insights in your data that can inform your modeling approach and
lead to better results.

Solving Problems
By exploring your data visually and statistically, you can identify and address potential issues or challenges before
they impact your modeling efforts.
Conclusion
Understand Data
EDA helps you thoroughly comprehend the data you're working with,
laying the foundation for successful machine learning projects.

Identify Insights
By exploring your data visually and statistically, you can uncover
valuable insights that can inform your modeling and decision-making.

Enhance Models
EDA enables you to make informed decisions that will enhance the
performance and reliability of your machine learning models.
Embrace the Power of
EDA
Understand Identify Solve Inform
Data Patterns Problems Decisions

Get familiar Uncover Address Make well-


with your trends, potential informed
dataset and relationship issues or decisions
its s, and challenges throughout
characteristi insights in before they the data
cs. your data. impact your science
modeling. workflow.

You might also like