NeoLung: Lung cancer prediction using machine learning

Aim:

The purpose of this project is to comapare Classification algorithms implemented on Lung Cancer Dataset

Dataset:

The Lung cancer dataset used in the project has been collected from data.world whose link is:

https://data.world/sta427ceyin/survey-lung-cancer

Working:

We have selected 10 of the following classification algorithms that have been used in this project:

• Logistic Regression
• K-Nearest Neighbors (KNN)
• Decision Tree
• Support Vector Machines (SVM)
• Naive Bayes
• Random Forest
• Gradient Boosting
• Neural Networks
• AdaBoost
• XGBoost

Then we build the model for each of the above mentioned algorithms. Using the following Evaluation Metrics we have compared the algorithms:

• Accuracy
• Precision
• F1 Score
• Recall Score
• Confusion Matrix

These are the accuracies of the algorithms:

• Logistic Regression: 90.29%
• K-Nearest Neighbors (KNN): 87.37%
• Decision Tree: 87.37%
• Support Vector Machines (SVM): 84.46%
• Naive Bayes: 86.4%
• Random Forest: 89.32%
• Gradient Boosting: 89.32%
• Neural Networks: 84.46%
• AdaBoost: 84.46%
• XGBoost: 84.46%

Results:

Out of all the algorithms so implemented, Logistic Regression performed the best. The evaluation metrics for Logistic Regression is as follows:

Accuracy: 0.9029126213592233

Precision: 0.9052631578947369

Recall: 0.9885057471264368

F1 score: 0.945054945054945

Confusion Matrix: