dmusili / creditcardfrauddetection Goto Github PK

Data Insight presents a specialized solution to these challenges with a strategic focus on effective fraud detection: Advanced Analytics for Large-Scale Data: Our model is meticulously designed to thoroughly analyze vast datasets, ensuring comprehensive fraud detection without compromising on quality.

Innovative Techniques for Imbalanced Data: We employ sophisticated methods to address data imbalance, thereby enhancing the detection of fraudulent transactions amidst a sea of legitimate activity.

Adaptable Model: The strength of our model lies in its simplicity. It is built to adapt seamlessly to the changing patterns of fraudulent behavior, allowing for quick updates and redeployment of the detection system as needed.

Entrust your bank's fraud detection to Data Insight, where we prioritize effective, precise, and privacy-conscious solutions to protect your customers and maintain the highest standard of financial security.

Source file creditcard.cvs (downloaded from Kaggle)

Data Cleanup

1. We dropped the time from the X data and separated the target
2. We split the training and test data
3. We scaled the x data using MinMaxScaler
4. We oversampled the training data
5. We undersampled the testing data

Top Performing Models We determined that the models below would be the most ideal for Credit Card Fraud detection

1. Support Vector Classifier
2. ADA Boost Classifier
3. Gradient Boosting Classifier!

Winning Model Based on our analsysis, we determined that the winning model was Support Vector Classifier