Skip to content

Using Deep learning (Artificial Neural network ) we are detecting fraud transaction using anonymized data. As such data is highly biased (fraud transactions are less than 0.1 % in real life ) ,we are using smote to over-sample the data .

Notifications You must be signed in to change notification settings

VikasSinghBhadouria/Credit-Card-Fraud-Detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 

Repository files navigation

Credit-Card-Fraud-Detection

Using Deep learning (Artificial Neural network ) we are detecting fraud transaction using anonymized data. As such data is highly biased (fraud transactions are less than 0.1 % in real life ) , we are using smote to over-sample the data .

Libraries needed :

  1. Numpy
  2. Pandas
  3. Keras
  4. imblearn In cases of highly baised data such as bank transaction,one class (fraud transactions , in this case ) very few. As a result , Model is not well trained for detecting observation for that class . In such cases ,either we can use undersampling or Oversapmling. We chose Oversampling using Smote .

There are other ways to handle such datasets . Ensemble methods are found to good for handling imbalance data. Random forest is found to be good at handling imbalance data. In this code , you can observe that ,we are getting better accuracy and loss with out smote but , when we use the models to predict for whole dataset , we got better results with deeplearning +smote in detecting Fraud detection. Incase of any error/query /suggestion feel free to contact me at [email protected]

Refer : https://medium.com/coinmonks/smote-and-adasyn-handling-imbalanced-data-set-34f5223e167

About

Using Deep learning (Artificial Neural network ) we are detecting fraud transaction using anonymized data. As such data is highly biased (fraud transactions are less than 0.1 % in real life ) ,we are using smote to over-sample the data .

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published