Data Science & Analytics
Batch 2022/FALL
Project Report
PROJECT TITLE
Breast Cancer Prediction Using Machine Learning
Supervisor: Engr. Farhan Ahmad
DEPARTMENT OF COMPUTER SYSTEM ENGINEERING
DAWOOD UNIVERSITY OF ENGINEERING & TECHNOLOGY, KARACHI Page | 1
Data Science & Analytics
1. Introduction & Problem Statement
Breast cancer is one of the most dangerous diseases affecting women worldwide. Early detection
of breast cancer can significantly increase survival rates and reduce treatment costs. However,
traditional diagnostic methods often depend on manual testing and expert opinion, which may be
time-consuming and sometimes inaccurate.
With the rapid advancement of Data Science and Machine Learning, intelligent systems can now
assist doctors in making faster and more accurate decisions. This project focuses on building a
machine learning-based healthcare analytics system that predicts whether a breast tumor is
malignant (cancerous) or benign (non-cancerous) using patient medical data.
1.1 Problem Statement
Manual diagnosis is time-consuming
Human error can occur
Early-stage cancer is difficult to detect
Need for an automated and accurate prediction system
1.2 Objective:
To develop a data-driven predictive model that helps in the early detection of breast cancer with
high accuracy.
2. Data Acquisition & Preprocessing
The dataset used in this project is taken from Kaggle, known as the Breast Cancer Dataset. It
contains 569 patient records and 30 numerical medical features, such as radius, texture,
concavity, symmetry, etc. The target column has only two classes:
0 → Benign
1 → Malignant
2.1 Data Preprocessing Steps
Removed unnecessary column: Unnamed: 32
Checked and handled missing values
Used Mean Imputation to fill NaN values
Applied Feature Scaling (Standardization)
DEPARTMENT OF COMPUTER SYSTEM ENGINEERING
DAWOOD UNIVERSITY OF ENGINEERING & TECHNOLOGY, KARACHI Page | 2
Data Science & Analytics
Divided data into:
o 80% Training Data
o 20% Testing Data
Preprocessing was essential to ensure that the machine learning model works efficiently and
without errors.
3. Applied Algorithms & Methodology
For breast cancer prediction, Logistic Regression was used. It is a supervised machine learning
algorithm suitable for binary classification problems.
3.1 Principles of Logistic Regression
It predicts probabilities
Uses the sigmoid function
Works best for binary outcomes
Fast and computationally efficient
Easy to interpret in medical applications
3.2 Overall Methodology
Data Collection
Data Cleaning & Preprocessing
Exploratory Data Analysis
Model Training
Model Testing
Performance Evaluation
4. Experimental Results & Performance Evaluation
After training the Logistic Regression model, it was tested on unseen data.
DEPARTMENT OF COMPUTER SYSTEM ENGINEERING
DAWOOD UNIVERSITY OF ENGINEERING & TECHNOLOGY, KARACHI Page | 3
Data Science & Analytics
4.1 Final Results
Patient 1 → 0 → Benign (cancer free)
Patient 2 → 1 → Malignant (cancer)
Patient 3 → 1 → Malignant (cancer)
Patient 4 → 0 → Benign (cancer free)
Patient 5 → 0 → Benign (cancer free)
DEPARTMENT OF COMPUTER SYSTEM ENGINEERING
DAWOOD UNIVERSITY OF ENGINEERING & TECHNOLOGY, KARACHI Page | 4
Data Science & Analytics
4.2 Performance Metrics
Precision (Malignant): 0.97
Recall (Malignant): 0.91
F1-Score (Malignant): 0.94
These results show that the model performs extremely well with very few misclassifications.
5. Discussion of Findings & Limitations
The model successfully predicts breast cancer with high accuracy. This shows that even simple
machine learning models can provide reliable results when trained on clean data. The system can
be used as a decision-support tool for doctors.
5.1 Limitations
Dataset only contains numerical data
No medical images used
Tested only in a simulated environment
Real hospital deployment not done yet
6. Conclusion & Recommendations
This project proves that machine learning can be effectively used in the healthcare sector for
early cancer detection. The Logistic Regression model achieved excellent accuracy and
reliability. The system can reduce diagnostic time and help in saving human lives.
6.1 Future Recommendations
Use deep learning models
Add medical images (MRI, Mammograms)
Deploy as a web or mobile application
Integrate with hospital systems
DEPARTMENT OF COMPUTER SYSTEM ENGINEERING
DAWOOD UNIVERSITY OF ENGINEERING & TECHNOLOGY, KARACHI Page | 5