A Data Science/Machine Learning Project. According to Bolster , Global Fraud Index (as at June 2022) is at 10,183 and growing. This is high risk to businesses and customers transacting online. This indicates that traditional rules-based methods of detecting and combating fraud are fast becoming less effective. It becomes imperative for stakeholders to develop innovative means to make transacting online as safe as possible. Artificial intelligence provides viable and efficient solutions via Machine Learning models/algorithms. In this project, I trained a fraud detection model to predict online payment fraud using Blossom Bank PLC as case study. Blosssom Bank ( BB PLC) is a multinational financial services group, that offers retail and investment banking, pension management, assets management and payment services, headquartered in London, UK. Blossom Bank wants to build a machine learning model to predict online payment fraud. Here is the dataset used for this task. With this model, BB PLC will: Keep up with fast evolving technological threats and better prevent the loss of funds (profit) to fraudsters. Accurately detect and identify anomalies in managing online transactions done on its platforms which may go undetected using traditional rules-based methods. 3.Improve quality assurance thus retaining old customers and acquire new ones. This will increase credit/profit base. Improve its policy and decision making. Steps: 1.Loading necessary python libraries. Loading Dataset. Exploratory Data Analysis. Higlighting Relationships and insights. Data Transformation; Using resampling techniques to address Class-imbalace.. Feature Engineering. Model Training. Model Evaluation. Challenges: I encountered a number of challenges during coding which made me run into error reports. these were due to improper documentations, syntax, especially during feature engineering (one-hot encoding: 'fit.transform'). This aspect consumed most of my time I was able to solve these challenges by making extensive research and paying close attention to syntax. I was able to selve the encoding by using 'pd.get_dummies() and making some specifications in the methods.