Scribd
Upload
50 views2 pages

Business Lean Model

The document proposes a machine learning and data-driven solution to detect fraud in real-time using multiple data sources. It would use APIs to gather structured and unstructured customer data, streamline the data, and use a Bayesian model to calculate compound risk scores and detect anomalies. A visualization tool would provide customized recommendations to reduce fraudulent activities by 80% and improve quality of service while achieving rapid return on investment. The target customers are telecom companies and financial institutions suffering major revenue losses from fraud.

Uploaded by

Vinay Kushwaha
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
50 views2 pages

Business Lean Model

The document proposes a machine learning and data-driven solution to detect fraud in real-time using multiple data sources. It would use APIs to gather structured and unstructured customer data, streamline the data, and use a Bayesian model to calculate compound risk scores and detect anomalies. A visualization tool would provide customized recommendations to reduce fraudulent activities by 80% and improve quality of service while achieving rapid return on investment. The target customers are telecom companies and financial institutions suffering major revenue losses from fraud.

Uploaded by

Vinay Kushwaha
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

Machine

Data Visual
Learning and Fraud
Data managment Recommenda
Data Sources Data Access Business Detection
collection and tions &
Intellegence/ Rules Engines
structuring Presentation
Analytics

Problems: Gather various data points of customer like (CDRs, IPDR, Device app info, biometric, geo-location, social, stream data, payments info
etc.)

Advance behavior fraud prediction model

Visualization of compound risk scoring.

Existing Alternative: Existing companies using legacy system based on single data point, only monitors, No real time fraud detection &
prevention.

Solution: Use APIs for data gathering from different data sources.

Use open-source technology stack to streamline structured and Un-structured.

Use ML (Bayesian model) to calculate compound risk score.

Provide customized recommendation via visualization tool.

Key Metrics: Fraud prevention recommendation engine.

Takes regular feedback, validate & do changes in the model based on feedback.

Early Fraud detection engine.

List of future/current customers who may be vulnerable.

Unique Value Proposition: Cost Effective

Scalable

Reduce Fraudulent & hack able activities to >80%

Rapid RoI and Increase QoS.


High Level Concept: Diyotta, Humbug Telecom Labs

Unfair Advantage: Unique ML algorithm, fraud detection Rules & Anomaly detection engine.

Customer Segment: Digital Service Provide, Telecom companies, financial institutions, Credit card companies etc.

Early adopters: Telecommunication companies who suffers from various kinds of frauds and revenue leakages due to which losing billions of $
every year.

Channels: unique fraud detection platform to customers to increase their revenue leakage due to frauds.

Raise awareness of FDP.

Get regular feedback and enhance the platform.

Cost Structure: Capex: Hardware cost, behavior prediction model creation

Opex: At least 3-4 man power required to manage tool

Few software purchasing licensing cost, scalable due to open source technologies.

Revenue Stream: Subscription services

Predictive analytics model

License fees

FDP can implement to other industries as well.

Provide recommendations to save $ B revenue.

You might also like