Multi-class_Classification

January 8, 2026

1 Multi-class Classification
Estimated time needed: 30 mins
In this lab, you will learn the different strategies of Multi-class classification and implement the
same on a real-world dataset.

1.1 Objectives
After completing this lab you will be able to:
1. Understand the use of one-hot encoding for categorical variables.
2. Implement logistic regression for multi-class classification using One-vs-All (OvA) and
One-vs-One (OvO) strategies.
3. Evaluate model performance using appropriate metrics.

1.2 Import Necessary Libraries

First, to ensure the availability of the required libraries, execute the cell below.

[1]: !pip install numpy==2.2.0

!pip install pandas==2.2.3
!pip install scikit-learn==1.6.0
!pip install matplotlib==3.9.3
!pip install seaborn==0.13.2

Requirement already satisfied: numpy==2.2.0 in /opt/conda/lib/python3.12/site-

packages (2.2.0)
Requirement already satisfied: pandas==2.2.3 in /opt/conda/lib/python3.12/site-
packages (2.2.3)
Requirement already satisfied: numpy>=1.26.0 in /opt/conda/lib/python3.12/site-
packages (from pandas==2.2.3) (2.2.0)
Requirement already satisfied: python-dateutil>=2.8.2 in
/opt/conda/lib/python3.12/site-packages (from pandas==2.2.3) (2.9.0.post0)
Requirement already satisfied: pytz>=2020.1 in /opt/conda/lib/python3.12/site-
packages (from pandas==2.2.3) (2024.2)
Requirement already satisfied: tzdata>=2022.7 in /opt/conda/lib/python3.12/site-
packages (from pandas==2.2.3) (2025.3)
Requirement already satisfied: six>=1.5 in /opt/conda/lib/python3.12/site-
packages (from python-dateutil>=2.8.2->pandas==2.2.3) (1.17.0)

1
Requirement already satisfied: scikit-learn==1.6.0 in
/opt/conda/lib/python3.12/site-packages (1.6.0)
Requirement already satisfied: numpy>=1.19.5 in /opt/conda/lib/python3.12/site-
packages (from scikit-learn==1.6.0) (2.2.0)
Requirement already satisfied: scipy>=1.6.0 in /opt/conda/lib/python3.12/site-
packages (from scikit-learn==1.6.0) (1.16.3)
Requirement already satisfied: joblib>=1.2.0 in /opt/conda/lib/python3.12/site-
packages (from scikit-learn==1.6.0) (1.5.3)
Requirement already satisfied: threadpoolctl>=3.1.0 in
/opt/conda/lib/python3.12/site-packages (from scikit-learn==1.6.0) (3.6.0)
Requirement already satisfied: matplotlib==3.9.3 in
/opt/conda/lib/python3.12/site-packages (3.9.3)
Requirement already satisfied: contourpy>=1.0.1 in
/opt/conda/lib/python3.12/site-packages (from matplotlib==3.9.3) (1.3.3)
Requirement already satisfied: cycler>=0.10 in /opt/conda/lib/python3.12/site-
packages (from matplotlib==3.9.3) (0.12.1)
Requirement already satisfied: fonttools>=4.22.0 in
/opt/conda/lib/python3.12/site-packages (from matplotlib==3.9.3) (4.61.1)
Requirement already satisfied: kiwisolver>=1.3.1 in
/opt/conda/lib/python3.12/site-packages (from matplotlib==3.9.3) (1.4.9)
Requirement already satisfied: numpy>=1.23 in /opt/conda/lib/python3.12/site-
packages (from matplotlib==3.9.3) (2.2.0)
Requirement already satisfied: packaging>=20.0 in
/opt/conda/lib/python3.12/site-packages (from matplotlib==3.9.3) (24.2)
Requirement already satisfied: pillow>=8 in /opt/conda/lib/python3.12/site-
packages (from matplotlib==3.9.3) (12.1.0)
Requirement already satisfied: pyparsing>=2.3.1 in
/opt/conda/lib/python3.12/site-packages (from matplotlib==3.9.3) (3.3.1)
Requirement already satisfied: python-dateutil>=2.7 in
/opt/conda/lib/python3.12/site-packages (from matplotlib==3.9.3) (2.9.0.post0)
Requirement already satisfied: six>=1.5 in /opt/conda/lib/python3.12/site-
packages (from python-dateutil>=2.7->matplotlib==3.9.3) (1.17.0)
Requirement already satisfied: seaborn==0.13.2 in
/opt/conda/lib/python3.12/site-packages (0.13.2)
Requirement already satisfied: numpy!=1.24.0,>=1.20 in
/opt/conda/lib/python3.12/site-packages (from seaborn==0.13.2) (2.2.0)
Requirement already satisfied: pandas>=1.2 in /opt/conda/lib/python3.12/site-
packages (from seaborn==0.13.2) (2.2.3)
Requirement already satisfied: matplotlib!=3.6.1,>=3.4 in
/opt/conda/lib/python3.12/site-packages (from seaborn==0.13.2) (3.9.3)
Requirement already satisfied: contourpy>=1.0.1 in
/opt/conda/lib/python3.12/site-packages (from
matplotlib!=3.6.1,>=3.4->seaborn==0.13.2) (1.3.3)
Requirement already satisfied: cycler>=0.10 in /opt/conda/lib/python3.12/site-
packages (from matplotlib!=3.6.1,>=3.4->seaborn==0.13.2) (0.12.1)
Requirement already satisfied: fonttools>=4.22.0 in
/opt/conda/lib/python3.12/site-packages (from
matplotlib!=3.6.1,>=3.4->seaborn==0.13.2) (4.61.1)

2
 Requirement already satisfied: kiwisolver>=1.3.1 in
/opt/conda/lib/python3.12/site-packages (from
matplotlib!=3.6.1,>=3.4->seaborn==0.13.2) (1.4.9)
Requirement already satisfied: packaging>=20.0 in
/opt/conda/lib/python3.12/site-packages (from
matplotlib!=3.6.1,>=3.4->seaborn==0.13.2) (24.2)
Requirement already satisfied: pillow>=8 in /opt/conda/lib/python3.12/site-
packages (from matplotlib!=3.6.1,>=3.4->seaborn==0.13.2) (12.1.0)
Requirement already satisfied: pyparsing>=2.3.1 in
/opt/conda/lib/python3.12/site-packages (from
matplotlib!=3.6.1,>=3.4->seaborn==0.13.2) (3.3.1)
Requirement already satisfied: python-dateutil>=2.7 in
/opt/conda/lib/python3.12/site-packages (from
matplotlib!=3.6.1,>=3.4->seaborn==0.13.2) (2.9.0.post0)
Requirement already satisfied: pytz>=2020.1 in /opt/conda/lib/python3.12/site-
packages (from pandas>=1.2->seaborn==0.13.2) (2024.2)
Requirement already satisfied: tzdata>=2022.7 in /opt/conda/lib/python3.12/site-
packages (from pandas>=1.2->seaborn==0.13.2) (2025.3)
Requirement already satisfied: six>=1.5 in /opt/conda/lib/python3.12/site-
packages (from python-dateutil>=2.7->matplotlib!=3.6.1,>=3.4->seaborn==0.13.2)
(1.17.0)
Now, import the necessary libraries for data processing, model training, and evaluation.

[2]: import pandas as pd

import numpy as np
import [Link] as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from [Link] import OneHotEncoder, StandardScaler
from sklearn.linear_model import LogisticRegression
from [Link] import OneVsOneClassifier
from [Link] import accuracy_score

import warnings
[Link]('ignore')

1.3 About the dataset

The data set being used for this lab is the “Obesity Risk Prediction” data set publically available
on UCI Library under the CCA 4.0 license. The data set has 17 attributes in total along with 2,111
samples.
The attributes of the dataset are descibed below.
Variable Name
Type
Description

3
Gender
Categorical
Age
Continuous
Height
Continuous
Weight
Continuous
family_history_with_overweight
Binary
Has a family member suffered or suffers from overweight?
FAVC
Binary
Do you eat high caloric food frequently?
FCVC
Integer
Do you usually eat vegetables in your meals?
NCP
Continuous
How many main meals do you have daily?
CAEC
Categorical
Do you eat any food between meals?
SMOKE
Binary
Do you smoke?
CH2O
Continuous
How much water do you drink daily?
SCC
Binary
Do you monitor the calories you eat daily?

4
 FAF
Continuous
How often do you have physical activity?
TUE
Integer
How much time do you use technological devices such as cell phone, videogames, television, com-
puter and others?
CALC
Categorical
How often do you drink alcohol?
MTRANS
Categorical
Which transportation do you usually use?
NObeyesdad
Categorical
Obesity level

1.3.1 Load the dataset

Load the data set by executing the code cell below.

[3]: file_path = "[Link]

↪GkDzb7bWrtvGXdPOfk6CIg/[Link]"

data = pd.read_csv(file_path)
[Link]()

[3]: Gender Age Height Weight family_history_with_overweight FAVC FCVC \

0 Female 21.0 1.62 64.0 yes no 2.0
1 Female 21.0 1.52 56.0 yes no 3.0
2 Male 23.0 1.80 77.0 yes no 2.0
3 Male 27.0 1.80 87.0 no no 3.0
4 Male 22.0 1.78 89.8 no no 2.0

NCP CAEC SMOKE CH2O SCC FAF TUE CALC \

0 3.0 Sometimes no 2.0 no 0.0 1.0 no
1 3.0 Sometimes yes 3.0 yes 3.0 0.0 Sometimes
2 3.0 Sometimes no 2.0 no 2.0 1.0 Frequently
3 3.0 Sometimes no 2.0 no 2.0 0.0 Frequently
4 1.0 Sometimes no 2.0 no 0.0 0.0 Sometimes

MTRANS NObeyesdad

5
 0 Public_Transportation Normal_Weight
1 Public_Transportation Normal_Weight
2 Public_Transportation Normal_Weight
3 Walking Overweight_Level_I
4 Public_Transportation Overweight_Level_II

1.4 Exploratory Data Analysis

Visualize the distribution of the target variable to understand the class balance.

[4]: # Distribution of target variable

[Link](y='NObeyesdad', data=data)
[Link]('Distribution of Obesity Levels')
[Link]()

This shows that the dataset is fairly balanced and does not require any special attention in terms
of biased training.

1.4.1 Exercise 1
Check for null values, and display a summary of the dataset (use .info() and .describe() meth-
ods).

6
[5]: # your code here

print([Link]())
print([Link]())
print([Link])

<class '[Link]'>
RangeIndex: 2111 entries, 0 to 2110
Data columns (total 17 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 Gender 2111 non-null object
1 Age 2111 non-null float64
2 Height 2111 non-null float64
3 Weight 2111 non-null float64
4 family_history_with_overweight 2111 non-null object
5 FAVC 2111 non-null object
6 FCVC 2111 non-null float64
7 NCP 2111 non-null float64
8 CAEC 2111 non-null object
9 SMOKE 2111 non-null object
10 CH2O 2111 non-null float64
11 SCC 2111 non-null object
12 FAF 2111 non-null float64
13 TUE 2111 non-null float64
14 CALC 2111 non-null object
15 MTRANS 2111 non-null object
16 NObeyesdad 2111 non-null object
dtypes: float64(8), object(9)
memory usage: 280.5+ KB
None
Age Height Weight FCVC NCP \
count 2111.000000 2111.000000 2111.000000 2111.000000 2111.000000
mean 24.312600 1.701677 86.586058 2.419043 2.685628
std 6.345968 0.093305 26.191172 0.533927 0.778039
min 14.000000 1.450000 39.000000 1.000000 1.000000
25% 19.947192 1.630000 65.473343 2.000000 2.658738
50% 22.777890 1.700499 83.000000 2.385502 3.000000
75% 26.000000 1.768464 107.430682 3.000000 3.000000
max 61.000000 1.980000 173.000000 3.000000 4.000000

CH2O FAF TUE

count 2111.000000 2111.000000 2111.000000
mean 2.008011 1.010298 0.657866
std 0.612953 0.850592 0.608927
min 1.000000 0.000000 0.000000
25% 1.584812 0.124505 0.000000
50% 2.000000 1.000000 0.625350

7
 75% 2.477420 1.666678 1.000000
max 3.000000 3.000000 2.000000
(2111, 17)
Expected Output:
• Counts of null values for each column (likely zero for this dataset).
• Dataset info including column names, data types, and memory usage.
• Descriptive statistics for numerical columns.

1.5 Preprocessing the data

1.5.1 Feature scaling
Scale the numerical features to standardize their ranges for better model performance.

[6]: # Standardizing continuous numerical features

continuous_columns = data.select_dtypes(include=['float64']).[Link]()

scaler = StandardScaler()
scaled_features = scaler.fit_transform(data[continuous_columns])

# Converting to a DataFrame
scaled_df = [Link](scaled_features, columns=scaler.
↪get_feature_names_out(continuous_columns))

# Combining with the original dataset

scaled_data = [Link]([[Link](columns=continuous_columns), scaled_df],␣
↪axis=1)

Standardization of data is important to better define the decision boundaries between classes by
making sure that the feature variations are in similar scales. The data is now ready to be used for
training and testing.

1.5.2 One-hot encoding

Convert categorical variables into numerical format using one-hot encoding.

[7]: # Identifying categorical columns

categorical_columns = scaled_data.select_dtypes(include=['object']).columns.
↪tolist()

categorical_columns.remove('NObeyesdad') # Exclude target column

# Applying one-hot encoding

encoder = OneHotEncoder(sparse_output=False, drop='first')
encoded_features = encoder.fit_transform(scaled_data[categorical_columns])

# Converting to a DataFrame
encoded_df = [Link](encoded_features, columns=encoder.
↪get_feature_names_out(categorical_columns))

8
 # Combining with the original dataset
prepped_data = [Link]([scaled_data.drop(columns=categorical_columns),␣
↪encoded_df], axis=1)

You will observe that all the categorical variables have now been modified to one-hot encoded
features. This increases the overall number of fields to 24.

1.5.3 Encode the target variable

[8]: # Encoding the target variable

prepped_data['NObeyesdad'] = prepped_data['NObeyesdad'].astype('category').cat.
↪codes

prepped_data.head()

[8]: NObeyesdad Age Height Weight FCVC NCP CH2O \

0 1 -0.522124 -0.875589 -0.862558 -0.785019 0.404153 -0.013073
1 1 -0.522124 -1.947599 -1.168077 1.088342 0.404153 1.618759
2 1 -0.206889 1.054029 -0.366090 -0.785019 0.404153 -0.013073
3 5 0.423582 1.054029 0.015808 1.088342 0.404153 -0.013073
4 6 -0.364507 0.839627 0.122740 -0.785019 -2.167023 -0.013073

FAF TUE Gender_Male … CAEC_no SMOKE_yes SCC_yes \

0 -1.188039 0.561997 0.0 … 0.0 0.0 0.0
1 2.339750 -1.080625 0.0 … 0.0 1.0 1.0
2 1.163820 0.561997 1.0 … 0.0 0.0 0.0
3 1.163820 -1.080625 1.0 … 0.0 0.0 0.0
4 -1.188039 -1.080625 1.0 … 0.0 0.0 0.0

CALC_Frequently CALC_Sometimes CALC_no MTRANS_Bike MTRANS_Motorbike \

0 0.0 0.0 1.0 0.0 0.0
1 0.0 1.0 0.0 0.0 0.0
2 1.0 0.0 0.0 0.0 0.0
3 1.0 0.0 0.0 0.0 0.0
4 0.0 1.0 0.0 0.0 0.0

MTRANS_Public_Transportation MTRANS_Walking
0 1.0 0.0
1 1.0 0.0
2 1.0 0.0
3 0.0 1.0
4 1.0 0.0

[5 rows x 24 columns]

9
 1.5.4 Separate the input and target data

[9]: # Preparing final dataset

X = prepped_data.drop('NObeyesdad', axis=1)
y = prepped_data['NObeyesdad']

1.6 Model training and evaluation

1.6.1 Splitting the data set
Split the data into training and testing subsets.

[10]: # Splitting data

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2,␣
↪random_state=42, stratify=y)

1.6.2 Logistic Regression with One-vs-All

In the One-vs-All approach:
• The algorithm trains a single binary classifier for each class.
• Each classifier learns to distinguish a single class from all the others combined.
• If there are k classes, k classifiers are trained.
• During prediction, the algorithm evaluates all classifiers on each input, and selects the class
with the highest confidence score as the predicted class.

Advantages:
• Simpler and more efficient in terms of the number of classifiers (k)
• Easier to implement for algorithms that naturally provide confidence scores (e.g., logistic
regression, SVM).

Disadvantages:
• Classifiers may struggle with class imbalance since each binary classifier must distinguish
between one class and the rest.
• Requires the classifier to perform well even with highly imbalanced datasets, as the “all”
group typically contains more samples than the “one” class.
Train a logistic regression model using the One-vs-All strategy and evaluate its performance.

[11]: # Training logistic regression model using One-vs-All (default)

model_ova = LogisticRegression(multi_class='ovr', max_iter=1000)
model_ova.fit(X_train, y_train)

[11]: LogisticRegression(max_iter=1000, multi_class='ovr')

You can now evaluate the accuracy of the trained model as a measure of its performance on unseen
testing data.

10
[12]: # Predictions
y_pred_ova = model_ova.predict(X_test)

# Evaluation metrics for OvA

print("One-vs-All (OvA) Strategy")
print(f"Accuracy: {[Link](100*accuracy_score(y_test, y_pred_ova),2)}%")

One-vs-All (OvA) Strategy

Accuracy: 76.12%

1.6.3 Logistic Regression with OvO

In the One-vs-One approach: * The algorithm trains a binary classifier for every pair of classes in
the dataset. * If there are k classes, this results in 𝑘(𝑘 − 1)/2 classifiers. * Each classifier is trained
to distinguish between two specific classes, ignoring the rest. * During prediction, all classifiers are
used, and a “voting” mechanism decides the final class by selecting the class that wins the majority
of pairwise comparisons.

Advantages:
• Suitable for algorithms that are computationally expensive to train on many samples because
each binary classifier deals with a smaller dataset (only samples from two classes).
• Can be more accurate in some cases since classifiers focus on distinguishing between two
specific classes at a time.

Disadvantages:
• Computationally expensive for datasets with a large number of classes due to the large number
of classifiers required.
• May lead to ambiguous predictions if voting results in a tie.
Train a logistic regression model using the One-vs-One (OvO) strategy and evaluate its performance.

[13]: # Training logistic regression model using One-vs-One

model_ovo = OneVsOneClassifier(LogisticRegression(max_iter=1000))
model_ovo.fit(X_train, y_train)

[13]: OneVsOneClassifier(estimator=LogisticRegression(max_iter=1000))

Evaluate the accuracy of the trained model as a measure of its performance on unseen testing data.

[14]: # Predictions
y_pred_ovo = model_ovo.predict(X_test)

# Evaluation metrics for OvO

print("One-vs-One (OvO) Strategy")
print(f"Accuracy: {[Link](100*accuracy_score(y_test, y_pred_ovo),2)}%")

One-vs-One (OvO) Strategy

Accuracy: 92.2%

11
 1.6.4 Exercises
Q1. Experiment with different test sizes in the train_test_split method (e.g., 0.1, 0.3) and observe
the impact on model performance.

[15]: # your code here

for test_size in [0.1, 0.3]:
X_train, X_test, y_train, y_test = train_test_split(X, y,␣
↪test_size=test_size, random_state=42, stratify=y)

model_ova.fit(X_train, y_train)
y_pred = model_ova.predict(X_test)
print(f"Test Size: {test_size}")
print("Accuracy:", accuracy_score(y_test, y_pred))

Test Size: 0.1

Accuracy: 0.7594339622641509
Test Size: 0.3
Accuracy: 0.749211356466877
Click here for the solution
for test_size in [0.1, 0.3]:
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=test_size, random_state
model_ova.fit(X_train, y_train)
y_pred = model_ova.predict(X_test)
print(f"Test Size: {test_size}")
print("Accuracy:", accuracy_score(y_test, y_pred))
Q2. Plot a bar chart of feature importance using the coefficients from the One vs All logistic
regression model. Also try for the One vs One model.

[ ]: # your code he
feature_importance = [Link]([Link](model_ova.coef_), axis=0)
[Link]([Link], feature_importance)
[Link]("Feature Importance")
[Link]("Importance")
[Link]()

Click here for the solution

# Feature importance
feature_importance = [Link]([Link](model_ova.coef_), axis=0)
[Link]([Link], feature_importance)
[Link]("Feature Importance")
[Link]("Importance")
[Link]()

# For One vs One model

# Collect all coefficients from each underlying binary classifier
coefs = [Link]([est.coef_[0] for est in model_ovo.estimators_])

12
 # Now take the mean across all those classifiers
feature_importance = [Link]([Link](coefs), axis=0)

# Plot feature importance

[Link]([Link], feature_importance)
[Link]("Feature Importance (One-vs-One)")
[Link]("Importance")
[Link]()
Q3. Write a function obesity_risk_pipeline to automate the entire pipeline:
Loading and preprocessing the data
Training the model
Evaluating the model
The function should accept the file path and test set size as the input arguments.

[18]: def obesity_risk_pipeline(data_path, test_size=0.2):

# Load data
data = pd.read_csv(data_path)

# Standardizing continuous numerical features

continuous_columns = data.select_dtypes(include=['float64']).columns.
↪tolist()

scaler = StandardScaler()
scaled_features = scaler.fit_transform(data[continuous_columns])

# Converting to a DataFrame
scaled_df = [Link](scaled_features, columns=scaler.
↪get_feature_names_out(continuous_columns))

# Combining with the original dataset

scaled_data = [Link]([[Link](columns=continuous_columns), scaled_df],␣
↪axis=1)

# Identifying categorical columns

categorical_columns = scaled_data.select_dtypes(include=['object']).columns.
↪tolist()

categorical_columns.remove('NObeyesdad') # Exclude target column

# Applying one-hot encoding

encoder = OneHotEncoder(sparse_output=False, drop='first')
encoded_features = encoder.fit_transform(scaled_data[categorical_columns])

# Converting to a DataFrame
encoded_df = [Link](encoded_features, columns=encoder.
↪get_feature_names_out(categorical_columns))

13
 # Combining with the original dataset
prepped_data = [Link]([scaled_data.drop(columns=categorical_columns),␣
↪encoded_df], axis=1)

# Encoding the target variable

prepped_data['NObeyesdad'] = prepped_data['NObeyesdad'].astype('category').
↪[Link]

# Preparing final dataset

X = prepped_data.drop('NObeyesdad', axis=1)
y = prepped_data['NObeyesdad']

# Splitting data
X_train, X_test, y_train, y_test = train_test_split(X, y,␣
↪test_size=test_size, random_state=42, stratify=y)

# Training and evaluation

model = LogisticRegression(multi_class='multinomial', max_iter=1000)
[Link](X_train, y_train)
y_pred = [Link](X_test)
print("Accuracy:", accuracy_score(y_test, y_pred))

# Call the pipeline function with file_path

obesity_risk_pipeline(file_path, test_size=0.2)

Accuracy: 0.8794326241134752
Click here for the solution
def obesity_risk_pipeline(data_path, test_size=0.2):
# Load data
data = pd.read_csv(data_path)

# Standardizing continuous numerical features

continuous_columns = data.select_dtypes(include=['float64']).[Link]()
scaler = StandardScaler()
scaled_features = scaler.fit_transform(data[continuous_columns])

# Converting to a DataFrame
scaled_df = [Link](scaled_features, columns=scaler.get_feature_names_out(continuous_c

# Combining with the original dataset

scaled_data = [Link]([[Link](columns=continuous_columns), scaled_df], axis=1)

# Identifying categorical columns

categorical_columns = scaled_data.select_dtypes(include=['object']).[Link]()
categorical_columns.remove('NObeyesdad') # Exclude target column

14
 # Applying one-hot encoding
encoder = OneHotEncoder(sparse_output=False, drop='first')
encoded_features = encoder.fit_transform(scaled_data[categorical_columns])

# Converting to a DataFrame
encoded_df = [Link](encoded_features, columns=encoder.get_feature_names_out(categoric

# Combining with the original dataset

prepped_data = [Link]([scaled_data.drop(columns=categorical_columns), encoded_df], axis=

# Encoding the target variable

prepped_data['NObeyesdad'] = prepped_data['NObeyesdad'].astype('category').[Link]

# Preparing final dataset

X = prepped_data.drop('NObeyesdad', axis=1)
y = prepped_data['NObeyesdad']

# Splitting data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=test_size, random_state

# Training and evaluation

model = LogisticRegression(multi_class='multinomial', max_iter=1000)
[Link](X_train, y_train)
y_pred = [Link](X_test)
print("Accuracy:", accuracy_score(y_test, y_pred))

# Call the pipeline function with file_path

obesity_risk_pipeline(file_path, test_size=0.2)

1.6.5 Congratulations! You’re ready to move on to your next lesson!

1.7 Author
Abishek Gagneja
### Other Contributors
Jeff Grossman
<!– ## Changelog
Date | Version | Changed by | Change Description |
|:————|:——|:——————|:—————————————|
2024-11-05 | 1.0 Abhishek Gagnejan | Fresh version created |
2025-05-13 | 1.1 Anita Verma | Added the solution code for Ovo model Q2 |
© IBM Corporation. All rights reserved.

15