ESTHER AWINO- SCT213-C002-0089/2023

SECOND YEAR SECOND SEMESTER FOR BSC. DATA SCIENCE AND ANALYTICS

APRIL,2025

MACHINE LEARNING COMPREHENSIVE ASSIGNMENT 1-CLASSIFICATION

# Import required libraries

import pandas as pd

import numpy as np

import matplotlib.pyplot as plt

import seaborn as sns

from sklearn.model_selection import train_test_split

from sklearn.preprocessing import StandardScaler, LabelEncoder

from sklearn.neighbors import KNeighborsClassifier

from sklearn.tree import DecisionTreeClassifier

from sklearn.svm import SVC

from sklearn.linear_model import LogisticRegression

from sklearn.metrics import confusion_matrix, classification_report, accuracy_score

# Step 1 & 2: Load and describe dataset

url = 'https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/ML0120ENv3/
Dataset/ML0101EN_EDX_skill_up/cbb.csv'

df = pd.read_csv(url)

print("Dataset Shape:", df.shape)

print("\nFirst five rows:")

print(df.head())

print("\nData Info:")
print(df.info())

print("\nStatistical Summary:")

print(df.describe())

print("\nMissing Values:")

print(df.isnull().sum())

# Step 3: Data Preprocessing and Visualization

# Handle missing values (if any)

for col in df.columns:

if df[col].isnull().sum() > 0:

if df[col].dtype == 'object':

df[col] = df[col].fillna('None')

else:

df[col] = df[col].fillna(df[col].median())

# Encode 'CONF' categorical column

le = LabelEncoder()

df['CONF'] = le.fit_transform(df['CONF'])

# Target Variable: Did the team reach POSTSEASON (1) or not (0)

y = df['POSTSEASON'].apply(lambda x: 0 if x == 'None' else 1)

# Feature selection

features = ['ADJOE', 'ADJDE', 'BARTHAG', 'EFG_O', 'EFG_D', 'TOR', 'TORD',

'ORB', 'DRB', 'FTR', 'FTRD', '2P_O', '2P_D', '3P_O', '3P_D',

'ADJ_T', 'WAB', 'CONF']

X = df[features]
# Step 4: Data Visualization

# Create correlation heatmap

plt.figure(figsize=(12, 8))

correlation_matrix = df[['ADJOE', 'ADJDE', 'BARTHAG', 'EFG_O', 'EFG_D', 'TOR',

'TORD']].corr()

sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm', center=0)

plt.title('Correlation Heatmap of Key Features')

plt.tight_layout()

plt.show()

# Distribution of offensive and defensive efficiency

plt.figure(figsize=(12, 5))

plt.subplot(1, 2, 1)

sns.histplot(df['ADJOE'], kde=True)

plt.title('Distribution of Adjusted Offensive Efficiency')

plt.subplot(1, 2, 2)

sns.histplot(df['ADJDE'], kde=True)

plt.title('Distribution of Adjusted Defensive Efficiency')

plt.tight_layout()

plt.show()

print("Feature distributions and correlations visualized above.")

# Step 5: Normalization

scaler = StandardScaler()

X_scaled = scaler.fit_transform(X)

# Step 6: Training and Validation

X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.2, random_state=42)

models = {

'KNN': KNeighborsClassifier(n_neighbors=5),

'Decision Tree': DecisionTreeClassifier(random_state=42),

'SVM': SVC(kernel='rbf', random_state=42),

'Logistic Regression': LogisticRegression(random_state=42)

trained_models = {}

for name, model in models.items():

model.fit(X_train, y_train)

trained_models[name] = model

print(f"{name} model trained successfully.")

# Step 7: Model Evaluation

results = {}

plt.figure(figsize=(20,5))

for i, (name, model) in enumerate(trained_models.items(), 1):

plt.subplot(1, 4, i)

y_pred = model.predict(X_test)

cm = confusion_matrix(y_test, y_pred)
 sns.heatmap(cm, annot=True, fmt='d', cmap='Blues')

plt.title(f'{name} Confusion Matrix')

plt.xlabel('Predicted')

plt.ylabel('Actual')

acc = accuracy_score(y_test, y_pred)

results[name] = acc

plt.tight_layout()

plt.show()

for name, model in trained_models.items():

print(f"Classification Report for {name}:")

y_pred = model.predict(X_test)

print(classification_report(y_test, y_pred))

# Step 8: Comparative Analysis

print("\nComparative Accuracy:")

results_df = pd.DataFrame(list(results.items()), columns=['Model', 'Accuracy'])

print(results_df.sort_values(by='Accuracy', ascending=False))

# Conclusion

best_model = results_df.sort_values(by='Accuracy', ascending=False).iloc[0]

print(f"\nBest Performing Model: {best_model['Model']} with Accuracy: {best_model['Accuracy']:.2f}")

Results and Key insights.

 The dataset had missing values hence were handled by replacement with either null (for
categorical data) or medium (for numerical data ) for the dataset to remain interpretable and
maintain its central tendency.
 Preprocessing:
Categorical Encoding: The column 'CONF' was label-encoded.
Target Variable: Transformed the 'POSTSEASON' column into binary labels (1 for
postseason, 0 otherwise).
 Feature Selection: Key features such as 'ADJOE', 'ADJDE', 'BARTHAG', etc., were
extracted for analysis.

 Visualization:

From the Correlation Heatmap:

Strong Positive Correlations:
BARTHAG correlates positively with ADJOE and EFG_O.
EFG_O and ADJOE are positively linked, showing that better field goal percentages improve offensive
efficiency
Negative Correlations:
ADJDE negatively correlates with offensive metrics (lower values indicate better defense).
TOR negatively correlates with offensive metrics, meaning higher turnover rates reduce offensive
efficiency
Distribution plots
Offensive Efficiency (ADJOE)
Shows a normal distribution, indicating most teams hover near the league average with few extremes.
Defensive Efficiency (ADJDE)
Also follows a normal distribution but has a tighter spread , suggesting less variation across teams.

 Used StandardScaler for feature normalization to prepare data for training.

 Model training and validation:
The following models were used:

Classification Report for KNN:

precision recall f1-score support
 0 0.91 0.96 0.94 223
1 0.81 0.66 0.73 59

accuracy 0.90 282

macro avg 0.86 0.81 0.83 282
weighted avg 0.89 0.90 0.89 282

Classification Report for Decision Tree:

precision recall f1-score support

0 0.92 0.92 0.92 223

1 0.71 0.71 0.71 59

accuracy 0.88 282

macro avg 0.82 0.82 0.82 282
weighted avg 0.88 0.88 0.88 282

Classification Report for SVM:

precision recall f1-score support

0 0.93 0.97 0.95 223

1 0.88 0.71 0.79 59

accuracy 0.92 282

macro avg 0.90 0.84 0.87 282
weighted avg 0.92 0.92 0.92 282

Classification Report for Logistic Regression:

precision recall f1-score support

0 0.94 0.97 0.96 223

1 0.88 0.78 0.83 59

accuracy 0.93 282

macro avg 0.91 0.88 0.89 282
weighted avg 0.93 0.93 0.93 282

Comparative Accuracy:
Model Accuracy
3 Logistic Regression 0.932624
2 SVM 0.918440
0 KNN 0.897163
1 Decision Tree 0.879433

Best Performing Model: Logistic Regression with Accuracy: 0.93