Bank User Query Intent Classification Model
This model classifies queries into twelve distinct buckets:
- Sample Label: Category, Type, Subtype
- Label 0: DEPOSIT, SB/ CA/TERM DEPOSIT ACCOUNTS, Dispute in charges deducted
- Label 1: DEPOSIT, SB/ CA/TERM DEPOSIT ACCOUNTS, Minimum Balance Charges related
- Label 2: DIGITAL BANKING, ATM RELATED, Dispute in ATM AMC Charges
- Label 3: DIGITAL BANKING,FUND REMITTANCE: NEFT/ RTGS/ IMPS through Branch,Dispute in charges deducted
- Label 4: LOANS & ADVANCES, Education Loans, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
- Label 5: LOANS & ADVANCES, Govt Scheme loans, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
- Label 6: LOANS & ADVANCES, Home loans, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
- Label 7: LOANS & ADVANCES, OTHER ADVANCES, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
- Label 8: LOANS & ADVANCES, Personal loans, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
- Label 9: LOANS & ADVANCES, SME ADVANCES, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
- Label 10: LOANS & ADVANCES, VEHICLE LOANS, Discrepancy in Charges (Processing Fee/Documentation charges, Inspection charges, etc)
- Label 11: Others
Training Metrics
The following training metrics were observed over 10 epochs:
| Epoch | Loss | Accuracy | F1 Score |
|---|
| 1 | 0.4103 | 0.9182 | 0.9184 |
| 2 | 0.0672 | 0.9827 | 0.9828 |
| 3 | 0.0351 | 0.9917 | 0.9917 |
| 4 | 0.0221 | 0.9948 | 0.9948 |
| 5 | 0.0171 | 0.9942 | 0.9943 |
| 6 | 0.0107 | 0.9966 | 0.9966 |
| 7 | 0.0056 | 0.9989 | 0.9989 |
| 8 | 0.0037 | 0.9986 | 0.9986 |
| 9 | 0.0152 | 0.9955 | 0.9955 |
| 10 | 0.0061 | 0.9982 | 0.9982 |
