Scope Of Project :

Feature Extracting :

• Data preprocessing & Data Visualisation (how many classes are divided in our dataset ).
• Over Sample & Under Sample (if data is imbalance make it balance by level uping )
• Normalisation (making larger values into samller values ).

Training Algorithm :

• Divide data set into three parts :

1. Training set
2. Validation set
3. Testing set

Machine learning models and algorithms for fraud detection :

1. (Logistics Regression
2. Support Vector Classifier
3. Decision Tree
4. K-Nears Neighbors classifier
5. Single Layer classifier
6. Multi Layer Classifier

Results of above models :

• By using the logistic regression , I have got the accuracy for the test data set as 80%
• By using the K – Nearest Neighbors(KNN), I have got the accuracy for the test data set as 80% which is same as logistic regression
• By using the Multi Layer Classifier (MLP), I have got the accuracy for the test data set as 84.168% which is comparatively more than other models
• By using the Single layer Perceptron (SLP), I have got the accuracy for the test data set as 72.5% which is comparatively lesser than other models
• By using the Support Vector Classifier (SVC) , I have got the accuracy for the test data set as 80% which is same as log reg and KNN
• By using the Decision Tree Classifier , I have got the accuracy for the test data set as 83% which is comparatively less than MLP
• By Comparing above model’s accuracy the Best Model is “MLP Classifier”

ROC_SCORE and ROC_CURVE :

• Checking True Positive Rate and False Positive Rate by ploting at different rate

Confussion Matrix :

• Distingushing the performance of a given classifier
• Confusion Matrix gives us a comparison between actual and predicted values
• ACC = (TP + TN)/(TP + FP + FN + TN)

K- Fold Cross Validation :

• K-Fold Cross-Validation. Cross-validation is a resampling procedure used to evaluate machine learning models on a limited data sample
• The procedure has a single parameter called k that refers to the number of groups that a given data sample is to be split into

Results :

• https://drive.google.com/file/d/13kV3HgijZe0syQVhNWrEjH3SRLR5e6_N/view?usp=sharing