Due in large part to delayed diagnosis and the lack of early-stage symptoms, lung cancer continues to rank among the deadliest illnesses worldwide. Early identification is critical to improving treatment outcomes and survival rates. This study proposes a comprehensive machine learning framework for lung cancer prediction using symptom-based patient data. The dataset, sourced from Kaggle, contains 1000 patient records with 25 attributes including lifestyle and symptom indicators such as smoking habits, anxiety, coughing, and chest pain. Ten classification algorithms, including Logistic Regression, Decision Tree, K-Nearest Neighbors, Naive Bayes variants, Random Forest, Support Vector Classifier, Gradient Boosting, XGBoost, and Multi-Layer Perceptron were assessed using Python libraries Scikit-learn and XGBoost. Performance was evaluated via accuracy, precision, recall, F1-score, and computational time based on a 70:30 train-test split. Ensemble models, particularly Random Forest and XGBoost, achieved the highest accuracy, consistent with cited literature. Limitations include the small dataset size, absence of cross-validation, and lack of external validation, which may affect generalizability. Future work should explore larger, diverse datasets and incorporate real-time clinical data to enhance diagnostic reliability. This approach has potential for integration into clinical decision support tools for early lung cancer detection.