In this credit card fraud prediction project, we aimed to develop models capable of accurately identifying potentially fraudulent transactions. The dataset encompassed various attributes including transaction amount, gender, city population, age, transaction month and year, as well as geographical coordinates, which were used to train and evaluate the models. Two distinct machine learning algorithms, namely Logistic Regression and Decision Tree, were employed for this task. The models were fine-tuned using techniques such as feature selection and hyperparameter tuning to enhance their predictive accuracy. The Logistic Regression model demonstrated an accuracy of X% on the test set, showcasing its proficiency in distinguishing between fraudulent and nonfraudulent transactions. Additionally, the Decision Tree model exhibited a similar accuracy of Y%. Through an in-depth analysis of feature importance, it was revealed that attributes such as transaction amount, age, and latitudinal distance played pivotal roles in making accurate predictions. Furthermore, the project emphasizes the significance of employing soft computing techniques, which involve computational methods designed to emulate human cognitive processes, for credit card fraud detection. By leveraging machine learning algorithms and feature engineering, we were able to construct robust models capable of safeguarding financial transactions from potentially fraudulent activities. This project highlights the efficacy of modern data-driven approaches in enhancing security measures within the financial industry. Keywords— Intrusion detection, Machine learning, Optimization algorithms.