5.5. Exploratory Data Analysis with Matplotlib and Seaborn: Bivariate analysis (relationships between two variables)

5.5 Exploratory Data Analysis with Matplotlib and Seaborn: Bivariate Analysis

Bivariate analysis is a fundamental aspect of Exploratory Data Analysis (EDA) that focuses on investigating the relationships between two variables. This type of analysis allows you to understand how one variable can affect or be related to another. In Machine Learning and Deep Learning, it is crucial to identify these relationships for feature selection, feature engineering, and to improve model interpretation. Python, with its Matplotlib and Seaborn libraries, offers powerful tools for visualizing and interpreting these relationships.

Matplotlib is a plotting library for the Python programming language and its numerical extension, NumPy. It provides an object-oriented programming interface for embedding graphics in applications that use general-purpose user interface toolkits such as Tkinter, wxPython, Qt, or GTK. On the other hand, Seaborn is built on top of Matplotlib and offers a high-level interface for drawing more attractive and informative statistical plots.

Types of Bivariate Graphs

There are several types of graphs that can be used for bivariate analysis, depending on the type of data you have:

• Scatter Plot: Used to visualize the relationship between two continuous variables. The points on the graph represent the intersection of the values ​​of the variables on the X-axis and the Y-axis.
• Line Plot: Similar to the scatter plot, but the points are connected by lines. It is useful for visualizing data over time (time series).
• Bar Plot: Used to compare categorical variables with continuous variables. The bars represent the magnitude of the continuous variable for each category.
• Box Plot: Shows the distribution of quantitative data in a way that facilitates comparisons between variables or between levels of a categorical variable. The "whiskers" extend to points that are within the 1.5x interquartile range, and points outside this range are considered outliers.
• Heatmap: A color chart that shows the magnitude of a phenomenon as color in two dimensions. It is useful for visualizing correlation matrices between variables.

Bivariate Analysis with Matplotlib and Seaborn

To perform effective bivariate analysis, it is important to understand how to use Matplotlib and Seaborn to create graphs that reveal relationships between variables. Let's explore some practical examples:

Scatter Plot with Matplotlib

To create a scatterplot with Matplotlib, you can use the scatter():

function

import matplotlib.pyplot as plt

# Example data
x = [value_x1, value_x2, value_x3, ...]
y = [y_value1, y_value2, y_value3, ...]

# Creating the scatter plot
plt.scatter(x, y)
plt.title('Scatter Plot between X and Y')
plt.xlabel('Variable X')
plt.ylabel('Variable Y')
plt.show()

Bar Plot with Seaborn

For a bar chart, Seaborn offers the barplot() function, which simplifies creation and adds more functionality:

import seaborn as sns

# Example data
categories = ['Category 1', 'Category 2', 'Category 3']
values ​​= [value_1, value_2, value_3]

# Creating the bar chart
sns.barplot(x=categories, y=values)
plt.title('Value Bar Chart by Category')
plt.xlabel('Categories')
plt.ylabel('Values')
plt.show()

Box Plot with Seaborn

Seaborn makes creating box plots simple with the boxplot() function:

# Example data
data = df[['categorical_variable', 'continuous_variable']]

# Creating the box plot
sns.boxplot(x='categorical_variable', y='continuous_variable', data=data)
plt.title('Box Plot of Continuous Variable by Category')
plt.xlabel('Category')
plt.ylabel('Continuous Variable')
plt.show()

Correlation Analysis with Heatmap

To visualize the correlation between multiple continuous variables, you can use a heatmap to show the correlation matrix:

# Calculating the correlation matrix
corr = df.corr()

# Creating the heatmap
sns.heatmap(corr, annot=True, fmt=".2f", cmap='coolwarm')
plt.title('Correlation Matrix Heatmap')
plt.show()

Final Considerations

Bivariate analysis is an essential part of data exploration and can provide valuable insights into how variables interact with each other. Using Matplotlib and Seaborn to visualize these relationships helps make the analysis more intuitive and accessible. By understanding the relationship between two variables, it is possible to make decisionstions in building Machine Learning and Deep Learning models.

It is important to note that visualization is only one part of bivariate analysis. Other statistical techniques, such as calculating the Pearson or Spearman correlation coefficient, are also important for quantifying the strength and direction of relationships between variables.

With practice and application of these visualization techniques, you will become more effective at interpreting data and identifying patterns that can be crucial to the success of your machine learning projects.