TRAVEL INDUSTRY
- Almost 10% of global GDP can be attributed to travel and hospitality industry
- Fraudsters are naturally attracted to a sector with high monetary value, rapid growth and large amount of online and digital transactions
- Various travel frauds arise from fake travel agents, transaction with stolen cards and reselling in grey market, sham website, insurance fraud etc.
$5.7 trillion
spent in travel and
hospitality industry in
2021
TRAVEL INSURANCE FRAUD
- Fabricated trip cancellation when the real reason is not covered by insurance
- False claim for lost baggage
- Exaggerating the value of items in allegedly lost baggage
- Fictitious medical treatment supposedly received during a trip
~20%
of travel claims
are fraudulent
PREVENTION OF INSURANCE FRAUD
- Each insurance claim is scrutinized one by one and if possible, compare with the previous claim history of the insured.
- Artificial intelligence and machine learning tools are widely adopted which are trained over lots of data to look for anomalies of potentially fraudulent activity patterns. The methodologies used are
- Outlier detection
- Classification
- Clustering
- They consider a variety of features that a human may miss
CASE STUDY
- A Singapore based travel insurance company used to manually scrutinize each claim to detect potentially fraudulent ones
- The process involved pre specified rule as well as judgment
- Suspicious claims were subject to more detailed investigation
- The investigating practice involved many hours of manual work and inconsistence process
OBJECTIVE
- Focus on reduced number of claims and thus
- reduce the manual work in initial scrutiny
- while significantly improving fraud detection
CHALLENGES
- Data included 77,445 claim records of which only 120 had been determined to be potentially fraudulent
- Identified potentially fraudulent claims are rare events (0.15%) and therefore hard to detect
- It was however expected that there could be a large number of undetected fake claims
SOLUTION
- Gradient Boosting
- A powerful machine learning algorithm
- Each claim was scored reflecting the risk of fraud
- Higher the score higher was the probability of the claim being fraudulent
BENEFIT
- Process automation
- Bypass manual scrutiny
- Efficiency
- Save time and effort
- Increased accuracy
- Eliminate human error
- Cost saving
- reducing manual work + false claims