Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Fraud DetectionHernan Huwyler                      Madrid, Spain                  Fraud Risk Forum                      Ja...
Who detects fraud?  33%                 18%                                       14%             13%                     ...
legitimate                    recordlegitimate  record             legitimate               record     fraudulent record  ...
Objective:maximize correct predictions and maintain incorrect        predictions at an acceptable level
Data Analysis              Identify AnalyticsInternal  Data           Apply Analytics to DataIndustry  Data            Lea...
Data Analysis              1                Branch A        Branch C Invigilation   Branch B                           HQ ...
Data Cleansing Algorithms            Poor initial data              conversion        Factiva World Check                S...
Demos Vendors / Employees - Conflicts of interests Vendor Activity – Sequentiality and fetching Treasury – Unrecorded Paym...
Some techniques Statistical data modeling Data preprocessing Matching algorithms Peer group outliers and covariance Time-s...
Open discussions Supervised, unsupervised & hybrid techniques Quality data Design, implementation and evaluation Visualiza...
Upcoming SlideShare
Loading in …5
×

Fraud detection Presentation

1,415 views

Published on

Statistical data modeling
Data preprocessing
Matching algorithms
Peer group outliers and covariance
Time-series analysis

Published in: Economy & Finance
  • Be the first to comment

  • Be the first to like this

Fraud detection Presentation

  1. 1. Fraud DetectionHernan Huwyler Madrid, Spain Fraud Risk Forum January 2013
  2. 2. Who detects fraud? 33% 18% 14% 13% 6% • Fraud Risk • Internal Audit Management FRM • Suspicions• Tipoff • By —chance” Transaction Reporting Controls PwCs Global economic crime survey 2012
  3. 3. legitimate recordlegitimate record legitimate record fraudulent record out of 9 system flags
  4. 4. Objective:maximize correct predictions and maintain incorrect predictions at an acceptable level
  5. 5. Data Analysis Identify AnalyticsInternal Data Apply Analytics to DataIndustry Data Leads Refine
  6. 6. Data Analysis 1 Branch A Branch C Invigilation Branch B HQ Branch D Business Branch A BI Branch C Intiligence Branch B Branch D
  7. 7. Data Cleansing Algorithms Poor initial data conversion Factiva World Check System consolidations World Compliance Manual data entry Interfaces and Customized data baches cleansing rules
  8. 8. Demos Vendors / Employees - Conflicts of interests Vendor Activity – Sequentiality and fetching Treasury – Unrecorded Payments Treasury – Abnormalities and triangulations Several sources of data Combined attributes Normalization
  9. 9. Some techniques Statistical data modeling Data preprocessing Matching algorithms Peer group outliers and covariance Time-series analysis
  10. 10. Open discussions Supervised, unsupervised & hybrid techniques Quality data Design, implementation and evaluation Visualization tools E-business transactions Best practices

×