logisticregression.py

# -*- coding: utf-8 -*-
"""logisticRegression.ipynb

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1LvIxp2rzUgQognGg7XuatdDUQqBc0-hD
"""

import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, confusion_matrix, classification_report
import datasets

# data = {
#     'Age': [22, 25, 47, 52, 46, 56, 23, 39, 45, 35],
#     'Fare': [7.25, 71.83, 7.92, 8.05, 26.0, 83.16, 8.05, 31.28, 35.5, 8.05],
#     'Survived': [0, 1, 0, 0, 1, 1, 0, 1, 0, 0]  # Target variable (0 = No, 1 = Yes)
# }

!pip install datasets

from datasets import load_dataset
dataset = load_dataset('victor/titanic')

dataset

df1=dataset['train'].to_pandas()

df1

df1=df1.rename(columns={'2urvived':'Survived'})

X = df1[['Age', 'Fare']] #independent variables
y = df1['Survived'] #dependent variables

X

y

df2=dataset['test'].to_pandas()

df2=df2.rename(columns={'2urvived':'Survived'})

XX = df1[['Age', 'Fare']] #independent variables
yy = df1['Survived'] #dependent variables

# X_train, X_test, y_train, y_test = train_test_split(X,y,test_size=0.3, random_state=42)

X_train ,y_train = X,y

X_test, y_test = XX,yy

log_reg = LogisticRegression()

log_reg.fit(X_train, y_train)

y_pred = log_reg.predict(X_test)

#evaluate

print("Accuracy:", accuracy_score(y_test, y_pred))
print("Confusion Matrix:\n", confusion_matrix(y_test, y_pred))
print("Classification Report:\n", classification_report(y_test, y_pred))

"""The model predicts 73% overall correctly. Sounds okay at first glance, but accuracy is misleading because of imbalance (many more class 0 than class 1).

1. 945 True Negatives (TN) → correctly predicted class 0.
2. 22 False Positives (FP) → predicted 1 but actual was 0.
3. 327 False Negatives (FN) → predicted 0 but actual was 1.
4. 15 True Positives (TP) → correctly predicted class 1.

Class 0 (did not survive):
precision = 0.74 → 74% of predicted 0’s were correct  
recall = 0.98 → model found almost all actual 0’s  

Class 1 (survived):
precision = 0.41 → when it predicted survived, only 41% were truly survivors  
recall = 0.04 → it found **almost none** of the survivors  
f1-score = 0.08 → very poor balance for class 1  
"""

new_passenger = pd.DataFrame({'Age': [30], 'Fare': [100]})
prediction = log_reg.predict(new_passenger)
print("Prediction:", prediction[0])

"""⚠ Problem

The dataset is imbalanced: far more class 0 than class 1.

Logistic regression with only Age and Fare cannot separate survivors well.
"""

# Plot decision boundary
import matplotlib.pyplot as plt

x_min, x_max = X['Age'].min() - 5, X['Age'].max() + 5
y_min, y_max = X['Fare'].min() - 5, X['Fare'].max() + 5

xx, yy = np.meshgrid(np.linspace(x_min, x_max, 200),
                     np.linspace(y_min, y_max, 200))

Z = log_reg.predict(np.c_[xx.ravel(), yy.ravel()])
Z = Z.reshape(xx.shape)

plt.contourf(xx, yy, Z, alpha=0.3)
plt.scatter(X['Age'], X['Fare'], c=y, edgecolors='k', marker='o')
plt.xlabel("Age")
plt.ylabel("Fare")
plt.title("Logistic Regression Decision Boundary")
plt.show()

model = LogisticRegression(max_iter=1000, class_weight="balanced")
model.fit(X_train, y_train)

y_predd = model.predict(X_test)

print("Accuracy:", accuracy_score(y_test, y_predd))
print("\nConfusion Matrix:\n", confusion_matrix(y_test, y_predd))
print("\nClassification Report:\n", classification_report(y_test, y_predd, zero_division=0))

import seaborn as sns
cm = confusion_matrix(y_test, y_pred)
labels = [0, 1]

plt.figure(figsize=(5,4))
sns.heatmap(cm, annot=True, fmt='d', cmap="Blues", xticklabels=labels, yticklabels=labels)
plt.xlabel("Predicted")
plt.ylabel("Actual")
plt.title("Confusion Matrix Heatmap")
plt.show()