Interactive Features

In this chapter, we'll discover how to add user interactivity to our application using Streamlit widgets:

• ⚡ Sidebar for organizing widgets in a collapsible panel
• ⚡ Select box for choosing categorical features
• ⚡ Slider for adjusting numerical values
• ⚡ Radio buttons for making single-choice selections
• ⚡ Checkbox for toggle options

These widgets will allow users to dynamically modify model feature values, which will then update our model's predictions in real-time.

Build a Sidebar

Streamlit's sidebar provides a convenient way for users to interactively filter and select input features for our machine learning model, enhancing the app's functionality and user experience. To implement this, let's update the $TUTORIAL_HOME/streamlit_app.py file with the following code:

import streamlit as st

# import pandas to read the our data file
import pandas as pd

st.title("🤖 Machine Learning App")

st.write("Welcome to world of Machine Learning with Streamlit.")

with st.expander("Data"):
    st.write("**Raw Data**")
    # read the csv file
    df = pd.read_csv("data/penguins_cleaned.csv")
    df
    # define and display
    st.write("**X**")
    X = df.drop("species", axis=1)
    X

    st.write("**y**")
    y = df.species
    y

with st.expander("Data Visualization"):
    st.scatter_chart(
        df,
        x="bill_length_mm",
        y="body_mass_g",
        color="species",
    )

# Interactivity
# Columns:
# 'species', 'island', 'bill_length_mm', 'bill_depth_mm',
# 'flipper_length_mm', 'body_mass_g', 'sex'
with st.sidebar:
    st.header("Input Features")
    # Islands
    islands = df.island.unique().astype(str)
    island = st.selectbox(
        "Island",
        islands,
    )
    # Bill Length
    min, max, mean = (
        df.bill_length_mm.min(),
        df.bill_length_mm.max(),
        df.bill_length_mm.mean().round(2),
    )
    bill_length_mm = st.slider(
        "Bill Length(mm)",
        min_value=min,
        max_value=max,
        value=mean,
    )
    # Bill Depth
    min, max, mean = (
        df.bill_depth_mm.min(),
        df.bill_depth_mm.max(),
        df.bill_depth_mm.mean().round(2),
    )
    bill_depth_mm = st.slider(
        "Bill Depth(mm)",
        min_value=min,
        max_value=max,
        value=mean,
    )
    # Filpper Length
    min, max, mean = (
        df.flipper_length_mm.min().astype(float),
        df.flipper_length_mm.max().astype(float),
        df.flipper_length_mm.mean().round(2),
    )
    flipper_length_mm = st.slider(
        "Flipper Length(mm)",
        min_value=min,
        max_value=max,
        value=mean,
    )
    # Body Mass
    min, max, mean = (
        df.body_mass_g.min().astype(float),
        df.body_mass_g.max().astype(float),
        df.body_mass_g.mean().round(2),
    )
    body_mass_g = st.slider(
        "Body Mass(g)",
        min_value=min,
        max_value=max,
        value=mean,
    )
    # Gender
    gender = st.radio(
        "Gender",
        ("Male", "Female"),
    )

Feature Data Preprocessing

We will now filter the data DataFrame based on the selected features and use pd.concat, ensuring the new data undergoes the same preprocessing steps as our training data.

Edit and update the $TUTORIAL_HOME/streamlit_app.py with the following code,

import streamlit as st

# import pandas to read the our data file
import pandas as pd

st.title("🤖 Machine Learning App")

st.write("Welcome to world of Machine Learning with Streamlit.")

with st.expander("Data"):
    st.write("**Raw Data**")
    # read the csv file
    df = pd.read_csv("data/penguins_cleaned.csv")
    df
    # define and display
    st.write("**X**")
    x = df.drop("species", axis=1)
    x

    st.write("**y**")
    y = df.species
    y

with st.expander("Data Visualization"):
    st.scatter_chart(
        df,
        x="bill_length_mm",
        y="body_mass_g",
        color="species",
    )

# Interactivity
# Columns:
# 'species', 'island', 'bill_length_mm', 'bill_depth_mm',
# 'flipper_length_mm', 'body_mass_g', 'sex'
with st.sidebar:
    st.header("Input Features")
    # Islands
    islands = df.island.unique().astype(str)
    island = st.selectbox(
        "Island",
        islands,
    )
    # Bill Length
    min, max, mean = (
        df.bill_length_mm.min(),
        df.bill_length_mm.max(),
        df.bill_length_mm.mean().round(2),
    )
    bill_length_mm = st.slider(
        "Bill Length(mm)",
        min_value=min,
        max_value=max,
        value=mean,
    )
    # Bill Depth
    min, max, mean = (
        df.bill_depth_mm.min(),
        df.bill_depth_mm.max(),
        df.bill_depth_mm.mean().round(2),
    )
    bill_depth_mm = st.slider(
        "Bill Depth(mm)",
        min_value=min,
        max_value=max,
        value=mean,
    )
    # Filpper Length
    min, max, mean = (
        df.flipper_length_mm.min().astype(float),
        df.flipper_length_mm.max().astype(float),
        df.flipper_length_mm.mean().round(2),
    )
    flipper_length_mm = st.slider(
        "Flipper Length(mm)",
        min_value=min,
        max_value=max,
        value=mean,
    )
    # Body Mass
    min, max, mean = (
        df.body_mass_g.min().astype(float),
        df.body_mass_g.max().astype(float),
        df.body_mass_g.mean().round(2),
    )
    body_mass_g = st.slider(
        "Body Mass(g)",
        min_value=min,
        max_value=max,
        value=mean,
    )
    # Gender
    gender = st.radio(
        "Gender",
        ("male", "female"),
    )

# Dataframes for Input features
data = {
    "island": island,
    "bill_length_mm": bill_length_mm,
    "bill_depth_mm": bill_depth_mm,
    "flipper_length_mm": flipper_length_mm,
    "body_mass_g": body_mass_g,
    "sex": gender,
}
input_df = pd.DataFrame(data, index=[0])
input_penguins = pd.concat([input_df, X], axis=0)

with st.expander("Input Features"):
    st.write("**Input Penguins**")
    input_df
    st.write("**Combined Penguins Data**")
    input_penguins

Now that we've prepared our input data, let's handle the categorical variables using encoding techniques. Feature encoding is crucial for converting categorical data into a format suitable for machine learning models.

Helpful resources for encoding:

• Scikit-learn encoding guide: https://scikit-learn.org/stable/modules/preprocessing.html#encoding-categorical-features
• Pandas get_dummies documentation: https://pandas.pydata.org/docs/reference/api/pandas.get_dummies.html