Mastering BI Visualizations: Plotly, Matplotlib, and More in Python#

Why Visualization Matters#

• Simplifies Complexity: High-level decision-makers often lack the time to sift through raw data. Visuals reduce noise and highlight what’s important.
• Highlights Trends and Patterns: Outliers, trends, seasonal effects, correlations—these are more easily spotted in charts and graphs.
• Facilitates Faster Decision-Making: Quick insights lead to faster decisions, improving overall business agility.

Role of Python in BI Visualization#

Python has emerged as a powerhouse in data science and analytics. Its ecosystem boasts strong capabilities in data manipulation (pandas, NumPy) and machine learning (scikit-learn, TensorFlow). On top of these, Python’s visualization libraries provide the necessary tools for turning processed data into compelling visuals.

2.1 Installing and Importing Matplotlib#

Matplotlib can be installed via pip or conda:

1
pip install matplotlib

or

1
conda install matplotlib

Importing Matplotlib in your script typically looks like:

1
import matplotlib.pyplot as plt

This plt alias is standard within the Python community.

2.2 Plotting Your First Graph#

Once installed, creating a simple line chart is straightforward:

1
import matplotlib.pyplot as plt
2

3
# Sample Data
4
x = [1, 2, 3, 4, 5]
5
y = [2, 4, 6, 8, 10]
6

7
# Basic Line Plot
8
plt.plot(x, y)
9
plt.title('Basic Line Plot')
10
plt.xlabel('X Axis')
11
plt.ylabel('Y Axis')
12
plt.show()

This example generates a basic line plot of a linear function.

2.3 Customizing Plots#

Matplotlib allows you to change colors, line styles, markers, and more. For instance, you can modify the previous plot with these parameters:

1
plt.plot(x, y, color='green', linestyle='--', marker='o', label='Line 1')
2
plt.title('Customized Line Plot')
3
plt.xlabel('X Axis')
4
plt.ylabel('Y Axis')
5
plt.legend()
6
plt.grid(True)  # Adds a grid

You can also configure the size of the figure and the resolution:

1
plt.figure(figsize=(8, 4), dpi=100)
2
plt.plot(x, y, color='blue')
3
plt.show()

2.4 Subplots and Layout Management#

Creating multiple plots within a single figure is common in BI reporting. This is where subplots come into play.

1
# Two subplots in one row
2
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 4))
3

4
# Subplot 1
5
ax1.plot(x, y, color='red')
6
ax1.set_title('First Subplot')
7

8
# Subplot 2
9
ax2.bar(x, y, color='green')
10
ax2.set_title('Second Subplot')
11

12
plt.tight_layout()  # Avoid overlapping text
13
plt.show()

This code places two plots side by side. plt.subplots returns a figure and axes object(s), allowing you full control over each subplot. With these fundamentals, you can build a variety of charts like bar charts, histograms, scatter plots, and more.

3.1 Working with Styles and Themes#

Matplotlib has a set of built-in styles, and you can create custom styles:

1
plt.style.available  # List of available styles
2
plt.style.use('ggplot')  # Use ggplot style

You can craft your organization’s custom style for consistent branding:

1
plt.style.use('default')  # reset style

If you create a file named my_custom_style.mplstyle and include your chosen settings, you can load the style with:

1
plt.style.use('my_custom_style.mplstyle')

3.2 Annotations and Text#

Annotations are particularly useful in BI when you want to draw attention to specific data points.

1
x_point = 3
2
y_point = 6
3
plt.plot(x, y, marker='o')
4
plt.annotate(
5
    'Key Data Point',
6
    xy=(x_point, y_point),
7
    xytext=(x_point+0.5, y_point+1),
8
    arrowprops=dict(facecolor='black', shrink=0.05)
9
)
10
plt.show()

This snippet draws an arrow pointing to the data point at (3, 6) and adds text next to it.

3.3 3D Plots#

While 2D plots suffice for most BI tasks, 3D plots can be invaluable for visualizing multi-dimensional data or adding depth to presentations.

1
from mpl_toolkits.mplot3d import Axes3D
2
import numpy as np
3

4
fig = plt.figure()
5
ax = fig.add_subplot(111, projection='3d')
6

7
# Create sample 3D data
8
X = np.linspace(-5, 5, 50)
9
Y = np.linspace(-5, 5, 50)
10
X, Y = np.meshgrid(X, Y)
11
Z = np.sin(np.sqrt(X**2 + Y**2))
12

13
# Plot a 3D surface
14
surf = ax.plot_surface(X, Y, Z, cmap='viridis')
15
fig.colorbar(surf, shrink=0.5, aspect=5)
16

17
plt.title('3D Surface Plot')
18
plt.show()

Here, you generate a sinusoidal surface in 3D. Though 3D plots can be visually striking, ensure that they provide real insight in a BI context rather than purely aesthetic complexity.

3.4 Saving Your Figures#

Sharing visualizations is essential in BI. Matplotlib allows you to save figures in multiple formats:

1
plt.savefig('my_figure.png', dpi=300)

Feel free to save in PNG, JPG, PDF, or SVG depending on your needs.

4.1 Seaborn#

Seaborn is built on top of Matplotlib. It excels in statistical plots and comes with smart defaults:

1
import seaborn as sns
2

3
tips = sns.load_dataset('tips')
4
sns.barplot(x='day', y='total_bill', data=tips)
5
plt.title('Average Total Bill per Day')
6
plt.show()

Seaborn’s higher-level interface can save time for routine tasks. It also includes specialized plots like violin plots, pair plots, and heatmaps.

4.2 Altair#

Altair is a declarative visualization library that integrates well with Jupyter notebooks. It’s built around the Grammar of Graphics, making it intuitive for those familiar with R’s ggplot2:

1
import altair as alt
2
from vega_datasets import data
3

4
cars = data.cars()
5
chart = alt.Chart(cars).mark_circle().encode(
6
    x='Horsepower',
7
    y='Miles_per_Gallon',
8
    color='Origin'
9
)
10
chart

Altair emphasizes clarity and minimal code for common tasks.

5.1 Getting Started with Plotly#

Plotly can be installed via pip:

1
pip install plotly

Imports differ slightly based on usage:

1
import plotly.graph_objs as go
2
from plotly.offline import plot

Alternatively, you can use Plotly Express, a simplified API introduced for convenience:

1
import plotly.express as px

5.2 Basic Charts (Line, Bar, Scatter)#

Using Plotly Express, you can create a basic scatter plot:

1
import plotly.express as px
2

3
df = px.data.iris()
4
fig = px.scatter(df, x='sepal_width', y='sepal_length', color='species')
5
fig.show()

Note how the color parameter automatically applies different colors for each species—no need to manually assign color maps.

Bar Chart Example:

1
df = px.data.tips()
2
fig = px.bar(df, x='day', y='total_bill', color='sex', barmode='group')
3
fig.update_layout(title='Total Bills per Day Groups by Gender')
4
fig.show()

This snippet groups bars by day, overlaying gender categories side by side, which can be essential in BI when comparing subgroups.

5.3 Advanced Charts (3D, Maps, Gantt)#

1
import plotly.express as px
2

3
df_3d = px.data.iris()
4
fig = px.scatter_3d(
5
    df_3d,
6
    x='sepal_length',
7
    y='sepal_width',
8
    z='petal_length',
9
    color='species'
10
)
11
fig.update_layout(title='3D Scatter Plot of Iris Dataset')
12
fig.show()

1
df_map = px.data.gapminder().query("year==2007")
2
fig = px.choropleth(
3
    df_map,
4
    locations='iso_alpha',
5
    color='lifeExp',
6
    hover_name='country',
7
    color_continuous_scale=px.colors.sequential.Plasma
8
)
9
fig.update_layout(title='Life Expectancy by Country (2007)')
10
fig.show()

1
import plotly.figure_factory as ff
2

3
tasks = [
4
    dict(Task="Task A", Start='2020-01-01', Finish='2020-01-03'),
5
    dict(Task="Task B", Start='2020-01-02', Finish='2020-01-04')
6
]
7
fig = ff.create_gantt(tasks, index_col='Task', title='Project Timeline')
8
fig.show()

5.4 Interactive Dashboards with Dash#

Plotly’s dash framework lets you create web-based dashboards entirely in Python. This is especially powerful for BI applications where you need to create interactive, sharable analytics tools.

Example minimal Dash application:

1
import dash
2
from dash import dcc, html
3
import plotly.express as px
4
import pandas as pd
5

6
app = dash.Dash(__name__)
7
df = px.data.iris()
8

9
app.layout = html.Div([
10
    html.H1("Iris Dashboard"),
11
    dcc.Dropdown(
12
        id='species-filter',
13
        options=[{'label': s, 'value': s} for s in df['species'].unique()],
14
        value='setosa'
15
    ),
16
    dcc.Graph(id='iris-graph')
17
])
18

19
@app.callback(
20
    dash.dependencies.Output('iris-graph', 'figure'),
21
    [dash.dependencies.Input('species-filter', 'value')]
22
)
23
def update_graph(selected_species):
24
    filtered_df = df[df['species'] == selected_species]
25
    fig = px.scatter(filtered_df, x='sepal_width', y='sepal_length')
26
    return fig
27

28
if __name__ == '__main__':
29
    app.run_server(debug=True)

When you run python dash_app.py, Dash launches a local web server. Accessing the provided URL in your browser displays an interactive dashboard. This can be deployed to cloud platforms and shared across teams.

Hybrid Approaches#

• Jupyter Notebooks and nbconvert: Create inline Plotly or Matplotlib charts and export them to PDFs or HTML for distribution.
• Application Integrations: Libraries like Bokeh and Dash can embed interactive charts into websites or internal portals.

Integrations with frameworks like Flask or Django can turn your visualizations into fully-fledged data-driven applications.