Skip to content

Data Preparation

Before applying our machine learning model, we need to convert categorical variables into numerical format using One-Hot1 Encoding. In our penguin dataset, we'll use pandas get_dummies()2 to encode:

Features (X):

  • island - Categorical location of penguin
  • sex - Gender of penguin

Target (y):

  • species - Type of penguin (our prediction target)

📝 One-Hot Encoding converts categorical variables into binary (0 or 1) format. For example: 

# Original: island = ['Torgersen', 'Biscoe']
# After encoding:
# island_Torgersen = [1, 0]
# island_Biscoe    = [0, 1]

Encoding Features and Target

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

streamlit_app.py
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
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_raw = df.drop("species", axis=1)
    X_raw

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

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,
    )
    # Flipper 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_raw], axis=0)

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

## Data Preparation

## Encode X
X_encode = ["island", "sex"]
df_penguins = pd.get_dummies(input_penguins, prefix=X_encode)
X=df_penguins[1:]
input_row = df_penguins[:1]

## Encode Y
target_mapper = {
    "Adelie": 0,
    "Chinstrap": 1,
    "Gentoo": 2,
}


def target_encoder(val_y: str) -> int:
    return target_mapper[val_y]


y = y_raw.apply(target_encoder)

with st.expander("Data Preparation"):
    st.write("**Encoded X (input penguins)**")
    input_row
    st.write("**Encoded y**")
    y

After successfully preprocessing our penguin dataset with appropriate encoding and feature selection, let's move forward to training our model and calculating species prediction probabilities. This step will prepare us for creating interactive visualizations in Streamlit.

  1. https://machinelearningmastery.com/why-one-hot-encode-data-in-machine-learning/ ↩

  2. https://pandas.pydata.org/docs/reference/api/pandas.get_dummies.html ↩