Dashboard¶

Coming Home

After the polars safely return to their harbor, they report to their elders, as tradition requires. Aarla and her crew visit the long house of Ursa Maior, the polar matriarch. Handing out a round of cold ales, she addresses the expedition:

"Now, captain Aarla, what did you find out there? Make it a good story that the Skaldi will sing of for a long time. And we want some plots and pictures, too! We are scientists after all, not some drunk barbarians, HAR HAR HAR!"

Aarla takes a deep breath, and starts to tell:

"We were sailing the stormy seas, with our hearts full of bravery, until one day we discovered this: ..."

Create a visual report about the penguins.

Install streamlit¶

First, install the streamlit library:

pip install streamlit

Start the server¶

Create a new file dashboard.py:

import streamlit as st

st.write("here is our report ...")

Then, open a terminal in the folder with the code and run:

streamlit run dashboard.py

Open your browser at http://127.0.0.1:8501 . You should see the text from the program.

Format text with Markdown¶

The st.write() understands Markdown. Use it to add headings, enumerations and similar:

st.write("# Penguins and their Beaks")

Display the data¶

Load the penguin data and print the DataFrame:

import seaborn as sns

df = sns.load_dataset("penguins")
st.write(df)

You should see a nicely formatted table in the browser.

Streamlit plots¶

Streamlit has its own plotting functions for DataFrames. Here is a bar chart:

counts = df["species"].value_counts()
st.bar_chart(counts)

Here are a line chart and a scatter plot. All plots take very similar format options:

st.line_chart(data=df, x="body_mass_g", y=["bill_length_mm", "flipper_length_mm"],
            color=["#ff00ff", "#00ff00"],
            width=0, height=0,
            use_container_width=True)

st.scatter_chart(data=df, x="bill_length_mm", y="body_mass_g",
            color="species",
            width=0, height=500,
            use_container_width=True)

Matplotlib and Seaborn plots¶

Matplotlib and seaborn plots will be rendered as static images. Here is a histogram:

fig = plt.figure()
sns.histplot(df["body_mass_g"], kde=True)
st.pyplot(fig)

Tabs¶

As an example of many convenient HTML elements, here is how you would subdivide your dashboard into multiple tabs:

tab1, tab2, tab3 = st.tabs(["Part 1", "Part 2", "Part 3"])

 with tab1:
     st.write("## Part 1")
     ...

 with tab2:
     st.write("## Part 2")
     ...

Control elements¶

A very convenient feature is that you can make your dashboard interactive using one line functions:

Select one or multiple entries from a list of strings:

options = ["bill_length_mm", "bill_depth_mm", "flipper_length_mm"]

column: str = st.selectbox("pick a data column", options)

species: list[str] = st.multiselect("species", options=["Adelie", "Chinstrap", "Gentoo"])

Select a number between two boundaries:

min_val: int = st.slider("minimum value", 0, 100)

Input some text:

text: str = st.text_input("enter some text: ", value=":sunglasses:")

Switch parts of your dashboard on and off:

show_plots: bool = st.toggle('Show some plots')

Create an interactive dashboard by using the results of these functions in plotting or data filtering commands.

Caching¶

A streamlit script is fully executed every time the page is updated. With a larger dataset, you may want to cache the loading and/or data wrangling in a function:

@st.cache_data
def load_data():
    return sns.load_dataset('penguins')

df = load_data()