SlideShare a Scribd company logo
1 of 24
Download to read offline
H2O.ai
Insurance
Use Case Catalog
https://www.flipsnack.com/B889C8BBDC9/h2o-ai-insurance-use-case-catalog-_-insurance-webinar-april-20
22/full-view.html
H2O.ai Confidential
Democratizing
AI for
20,000
Organizations
6
OF THE TOP
INSURANCE
Companies
Partner with H2O.ai
v
H2O.ai Confidential
Infrastructure
and Operations
Pricing &
Underwriting
Customer 360 Customer Behavioral Analysis
Cyber-threat Detection
Employee Retention
Precision Pricing
Automated underwriting
Reserve Predictions
Claims
Management
Call Centers/case routing
Claims processing
Damage Detection
Estimation
H2O.ai Use Case Catalog for Insurance
Meeting the needs of Insurance Customers
Fraud Claims Fraud
Customer
Experience
KYC Automation
Targeting
Propensity to Buy
Churn prediction
Click on each for in depth coverage across Insurance use cases.
Infrastructure and Operations
v
H2O.ai Confidential
Cyber-threat detection
Objective
● Help information security teams reduce risk and improve their
security posture efficiently and effectively
● Automate threat detection across large numbers of devices,
attack vectors, and data silos
Outcome
● Continuous assessment of asset inventory to gain a complete and
accurate view of devices, users and applications with access to IT
systems
● Models that can detect and respond to deviations from the norm,
even with noisy data
● Prediction models that will assess where and how a company is
most likely to be breached, so planning and resource allocation
can be directed toward weak points in the IT system
● Explainability of model recommendations and analysis for
security operations leaders, CISOS, auditors and Board of
Directors
Business Value
● Get up to date knowledge of global and industry level threats to
help prioritize defense systems
● Prevent cyber threat incidents and respond quicker/better when
they do happen, improving OPEX
● Free up limited cybersecurity teams to focus on complicated
cases, while AI takes care of routine tasks
H2O.ai’s AI and Data Approaches
● Classification Models that can identify threatening vs non
threatening events and actors
● Anomaly detection, entry classification, domain generation
detection
● Unsupervised learning for unlabeled data, clustering data based
on anomalies
● Analyze large data sets of events to identify many different types
of threats (eg, malware, ransomware, email phishing, malicious
code downloads)
● Train neural networks to tell the difference between malicious
and safe files
● Use images to train classifier neural networks to detect malware
in .doc and .pdf files
● Bias reduction models
v
H2O.ai Confidential
Employee Retention
Objective
● Predict the risk of employee departure and identify relevant factors
● View predictions of employee departure, forecast churn rates and
identify relevant factors contained in employee data
Outcome
● Models with high accuracy of predicting employee departure and
identifying reasons for departure
Business Value
● Retain flight-risk talent and thereby improve per employee
efficiencies
● Minimize cost of employee replacement due to backfills
H2O’s AI and Data Approaches
● Automatic feature engineering, model validation, model tuning,
selection and deployment, machine learning interpretation (MLI),
recipes, time series and auto pipeline generation for model scoring
H2O AI App: Employee Churn Prediction
Pricing and Underwriting
v
H2O.ai Confidential
Precision Pricing
Objective
● Increase efficiency and accuracy of the entire pricing process
● React faster to rate changes with less actuarial and IT effort
● Improve model explainability and transparency for all lines of
business
Outcome
● Virtually eliminate model bias with automatic disparate impact
analysis for any model generated.
● Leverage new features gleaned from unstructured data through
OCR and NLP
● Reduce IT resources required to deploy and monitor models with
automated drift detection
● Automate the model documentation and filing process for state
regulators
Business Value
● Loss ratio improvement through enhanced segmentation and
targeted pricing
● Assure regulators and public that with unsurpassed model
explainability and interpretability
● Reduce risk and write more profitable, high quality business
H2O.ai AI and Data Approaches
● Automate the preparation of data pulled from from multiple
sources (eg, historical claims, policy and customer conversion and
retention data), with custom recipes to build models with the
insurance industry’s most widely used modeling platform
● Leverage automated feature engineering along with shapley
values for enhanced model interpretability and accuracy
● Provide product managers with “what if” pricing simulation AI
apps to predict impact of new pricing scenarios on profitability
and retention
v
H2O.ai Confidential
Automated Underwriting
Objective
● Increase efficiency and accuracy of risk pricing process associated
with traditional manual underwriting processes, increasing
competitive advantage for insurers
● Widen the scope of data sources underwriters can use for
evaluation
Outcome
● Automate underwriting process including data collection, data
extraction, and filling forms
● Extract information from unstructured data through OCR and NLP
● Make models that can learn from 3rd party data sources, claims
histories and other past data to predict risk profile of new
submissions
● Deeper visibility into customer risk profiles
Business Value
● Tailor premiums to match each individual’s actual risk and optimize
pricing
● Shorten underwriting workflows creating near real-time service for
customer experience
● Improve underwriting profitability, while reducing OPEX, customer
churn, and costs for customer retention
H2O .ai AI and Data Approaches
● Use H2O recipes to automate the leveraging of data from
multiple sources (historical policy data, policyholder churn &
claims data to build models with LightGBM, Random Forest,
XGBoost and neural networks that outperform traditional GLM
approaches
● Train models with explainability and interpretability; understand
variable importance
● Use deep learning to process images and raw machine data
v
H2O.ai Confidential
Reserve Prediction
Objective
● Use AI/Matching Learning to predict claim reserve amounts
based on historical based trends
● Accurately estimate the Reserve, and recalibrating these
estimates automatically when new information is added to the
claim record helps carriers react swiftly to ever changing events
Outcome
● Models that produce claims predictions that are aligned with actual
claims on an individual claim level, both in expected value and
variance.
Business Value
● Ensure that the size of reservoir of money Insurance Carriers hold is
adequately estimated the cost of future claims accurately, but not
excessive, to cover potential liabilities already assumed
● Improved transparency on future liabilities in order to ensure proper
statement of future liabilities on the balance sheet
H2O .ai AI and Data Approaches
● Machine learning techniques on a single, end-to-end platform
that leverages decision tree, SVM, neural network based and
deep learning based techniques
● Joint modeling of several numerical values (eg, paid losses and
claims outstanding)
● Make use of various claims data types (eg, numerical, categorical,
textual data) that can enhance prediction model understanding
and performance
Claims Management
v
H2O.ai Confidential
Customer Handling and Call Center Optimization
Objective
● Improve customer experience and reduce handling costs
● Map the query by the customer to appropriate resolution section
Outcome
● An improvised automated method of Exploratory Data Analysis that
segregates the data and provides numerical & graphical analysis
based on the factors such as Date/Time, Location, Self Made
Complaints, user profiles and past interactions.
● Improvised NLP techniques and real time processing can accurately
pinpoint/classify the complaints into groups.
● Send high risk, complex complaints to the more experienced
customer support agents.
● A multi-classification AI model to give suggestions for possible
resolution.
● The model can also be used as a backend mapping engine for
Chatbots and Customer Issue Resolution applications.
● The solution can be delivered through APIs and custom application
Business Value
● Address the complaints of customers, maintain trust and ultimately -
retain.
● Improve case deflection, the rate that customers are able to find
their own answers to issues that they would have otherwise called
support for.
H2O’s AI and Data Approaches
● Conduct Exploratory Data Analysis on Customer Complaint Data.
This data in the form of a Dataset will be Analysed, and numerical as
well as graphical analysis will be provided. Answers to questions
such as which location has the most complaints, what time period, is
the complaint made on behalf of the user or made by the customer,
etc. shall be provided in an organized manner.
● NLP Techniques will be used to process the data, which can be in
the form of a dataset or realtime.
● H2O AI App, built with H2O Wave, an SDK for Data Scientists:
○ Import Data and Data Prep ( Complaint Data)
○ Exploratory Data Analysis to analyse the aforementioned points.
○ Interactive Dashboard where insights are displayed based on the
searched tags, percentage rank of complaints ( #1 Rude
behaviour, #2 Issues, etc.) There will also be percentage
comparisons as the data increases ( Complaints in xxx domain
are 15% higher, etc.) Top increases, top decreases
○ Natural Language Processing Apps
● Applying NLP models to obtain tags based on meanings.
○ Text Sentiment Apps
v
H2O.ai Confidential
Claims Processing
Objective
● Infuse the claims processing workflow using AI to inform adjusters with
smart actions
● Quickly and accurately estimate and process claims
● Use AI to automate repetitive claims processes and streamline claims
processing, allowing those with low risk to be processed automatically
while higher risk claims are routed to investigators for review
● Predict claims payments at individual level
Outcome
● An improvised automated method of Exploratory Data Analysis that
segregates the data and provides numerical & graphical analysis based
on factors such as Date/Time, Location, Self Made Complaints, user
profiles and past interactions.
● An improvised automated method of Exploratory Data Analysis that
segregates the data and provides numerical & graphical analysis based
on factors such as Date/Time, Location, Self Made Complaints, user
profiles and past interactions.
● NLP techniques and real time processing can accurately
pinpoint/classify the complaints into groups.
● Send high risk, complex complaints to the more experienced customer
support agents.
● A multi-classification AI model to give suggestions for possible
resolution.
● The model can also be used as a backend mapping engine for Chatbots
and Customer Issue Resolution applications.
● The solution can be delivered through APIs and custom application
H2O’s AI and Data Approaches
● Build multiple models that can analyze all incoming structured and
unstructured data, scanned images, faxes and natural language sets,
policy information, medical bills reports, and legal documents to offer
real time recommendations to claims adjusters.
● Each time new claims data is added to a record, apply algorithms to
accurately score for fraud, reserve changes, and claim complexity at a
level of nuance that is often missed through manual reviews.
● Build models that can route high complexity claims to seasoned
adjuters for review, generating smart alerts with reason codes to take
immediate action; Low complexity, low fraud claims are automatically
flagged for straight through processing
Business Value
● Fewer touches
● Claims adjusters react fast to critical events
● Faster claims settlement
● Reduced ALAE
v
H2O.ai Confidential
Damage Detection and Estimation
Objective
● Increase granularity and accuracy for damage estimation
● Accurately detect and classify damages from images
Outcome
● Build models that can detect and classify damages from images,
isolating where in the image there is damage and what type of
damage is represented
● Translate damage into individual parts affected by adding additional
information to the analysis, such as parts listed needed to repair,
model optimum reconditioning scenarios
● Retrieve images of similar items (eg, vehicles), both damaged and
undamaged, to help estimators evaluate whether or not the item
was actually damaged
Business Value
● Shorten processing time for property and casualty damage claims,
improving customer satisfaction with faster, more accurate
settlements
● Equip claims estimators with signals to identify where to focus
attention during damage assessment, and reasons for it
● Route damage estimates to the right estimator team
H2O’s AI and Data Approaches
● H2O’s image recognition and deep learning capabilities that can
predict vehicle damage severity using multiclass classification
● Use binary classsification for vehicle damage identification
● Use multiclass classification of vehicle make, model, and year
● Use H2O Wave to make AI applications that allow business
stakeholders to get:
○ a pre-selection of cases where pictures do not match with
expectations
○ highlighted images where cases do not match claims reports
○ recommendations for next steps
● Explainable AI visualizations and narratives that explain why the
model made its damage estimation
Deliver on Customer Experience
v
H2O.ai Confidential
Targeting
Objective
● Send messages/content at the right time with the right content
based on individual and collective customer behaviors
● Help marketing teams at FIs achieve more awareness, higher
conversions and better engagements
Outcome
● Pool large data sets to accurately predict custom audience
segments with specific attributes relevant to the brand
● Identify unique and relevant audiences for the brand that will be
in line with the audience data pools initially created
Business Value
● Increase click through rates by 4x (industry expectations)
● Reduce costs per engagement
H2O’s AI and Data Approaches
● Machine learning interpretability (MLI) that provides targeting
explanations
● Ability to provide computational power to advance granular
audience targeting capabilities
● Target identification models - automate and optimize processes
for potential consumer identification, information extraction and
market segmentation
● Probability models that will predict level at which consumers will
click on ads, click through rates, etc
● Recommendation models - discover useful patterns to determine
what the user finds interesting or not
● Machine learning interpretability that provides targeting
explanations
H2O AI Apps, built with H2O Wave, an SDK for Data Scientists:
● UMAP-based clustering App
● Customer Journey / Lead Scoring App
v
H2O.ai Confidential
Objective
● Expand offers to customers with accuracy and speed by
implementing predictive analytics
● Forecast behaviors of a target audience by analyzing their past
behaviors
Outcome
● Cut time to build models from 6 months to less than a week
● Doubled customer propensity to buy rate
● Realized additional revenue by being better able to target offers
● Created propensity models that helped companies identify the
right customers and prospects that have a high likelihood to
purchase a particular product or service
Business Value
● Optimize promotions based on knowing which leads to sell to
with the right message and product/services
H2O’s AI and Data Approaches
● Create models and use custom recipes that generate
features/variables that provide a probabilistic estimation of
whether customers will perform any of such actions or not i.e. a
propensity score.
● Use propensity scores to estimate value each customer brings in
real-time. Make data available to Relationship Managers of the
Bank (RMOs) who can further slice and dice the data and
consume the information intuitively.
● H2O AI Apps, built with SDK for Data Scientists, can perform:
○ Data Preparation
○ Supervised Machine Learning
● Automatic feature engineering, model tuning and
optimization, scoring pipeline generation
● Accurate time series capabilities
● Automatically generated visualizations and data plots
○ Nonlinear algorithmic modeling
○ Binary classification
○ Results viewed on Dashboards in a tabular format, grouped by
different features (city, areacode, spending limit, etc) or for a
single user; view target groups and selected features such as
textual insights explaining the prediction, the propensity
score for each user (mean or median) if a target group is
selected.
Propensity to Buy
v
H2O.ai Confidential
Customer Churn
Objective
● Predict customer churn based on behavioral patterns and
cadence of transactions
● Identify leading indicators for churn, improving customer
retention insights
Outcome
● Optimized models up to the granularity of each customer
● Calculated daily probability for customer churn for early
detection
● Distributed in-memory infrastructure trains and scores entire
customer base in minutes instead of hours
● Reduced model building time from 6-7 hours to < 30 min
Business Value
● Intercepted customer churn before it happened
● Know what leads a client towards the decision to leave the
company.
● Develop loyalty programs and retention campaigns to keep as
many customers as possible.
● Improving the customer retention rate for existing customers
by just 5 percent can improve a company’s profitability by 25 to
95 percent (Bain & Co).
H2O’s AI and Data Approaches
● Identify customers with high probability of churning by creating models and
custom recipes specifically built for generating features/variables.
● Distributed algorithms (Random Forest, GBM)
● Capabilities to iterate through models with different parameters within time
constraints, instead of relying on just one model
● H2O Customer Churn AI App, built with H2O Wave:
Fraud and Compliance
v
H2O.ai Confidential
Claims Fraud
Objective
● Find new ways to improve fraud detection accuracy and
detection time across claims fraud types
● Use AI to generate new fraud rules
● Detect anomalous patterns for individuals, accounts and
networks within and across portfolios
● Target different groups and subsets in a tailored approach,
comparing different inferences from these subsets with the
same inferences from the entire population
● Flag specific repetitive fraud groups and deploy counter
measures
Outcome
● 6x faster development of state-of-the-art ML models that
pre-empted fraudulent claims
H2O’s AI and Data Approaches
● Build AI models and use custom recipes specifically built for generating
features/variables that provide associated information about fraudulent
behavior. This data is then available to the fraud investigator who can further
slice and dice the data and consume the information intuitively.
● Feature engineering with Deep Learning to model new and complex attack
patterns quickly
● Behavior profiling for data networks - IP addresses, buying patterns
● Terabytes of data leveraged to deliver high scalability and performance, flexible
deployment and integration with other big data frameworks
● Model Explainability aids investigators in understanding the why the model
thinks there is potential fraud.
Business Value
● Reduce fraud losses and improve customer
experience
● 2x increase in accurate claims fraud detection
● 11% improvement in accuracy resulting in $1M
saved monthly per basis point
Analyze the Customer 360
v
H2O.ai Confidential
Know your customer (KYC) Automation
Objective
● Operationalize customer data to understand behavior analytics for several use
cases across pricing, claims management, fraud prevention and more
● Identity verification
Outcome
● Create a foundation for customer analytics
● sA series of models that can assess customer analytics, behaviors and needs
across very different customer profiles
Business Value
● Improve campaign effectiveness
● Accurate, targeted cross-sell / up-sell
● Retain your most profitable customers
● Deliver real-time superior customer experience at all
points of service
● Increase customer lifetime value
● Verify that your customers are who they say they are
and assessing the risks associated with each customer
like the possibilities of fraud, money laundering,
terrorism financing, etc.
● Improve digital onboarding experience with automated
KYC processes, reducing application abandonment.
H2O’s AI and Data Approaches
● Process large quantities of customer data and integrate
into the H2O AI Cloud to build KYC models
● Probabilistic and Fuzzy Matching
● Document verification, NLP and image recognition to
verify and authenticate customer information
● Seamlessly extract customer data and integrate with
existing customer management or onboarding systems
Customer Data Integration
Social Profile Credit Profile Life Insurance
Profile
Mortgage Account
Profile
Bank Account
Profile
Customer
360º
View
Master Data Management
H2O AI Engine /
Probabilistic Matching
v
H2O.ai Confidential
Customer Behavior Analysis
Objective
● Segment users on the basis of different attributes such as -
demographics data, transaction history, transaction behavior
(Recency, Frequency, Monetary).
● Rapidly identify and evaluate customer behavior signals to take
action on potential fraudulent activity, optimize marketing
activities, etc
Outcomes
● Create behavior scoring models
● Continuously update customer data
Business Value
● Create foundation for:
○ Personalization strategies to improve customer experience,
customer sentiment
○ Targeting different groups in a tailored approach
○ Recommending core banking products for cross-selling and
up-selling, increase sales, revenues, and profits
○ Deploying customer churn prevention schemes
H2O’s AI, Data-Driven Approach:
● Provide an end-to-end pipeline that collects and updates the
customer data, performs data preparation, performs
unsupervised machine learning, and generates clustering results.
● Use H2O Wave (SDK for Data Scientists) to build an AI App that
can perform:
○ Data Preparation
○ Unsupervised Machine Learning
○ Rule fit analysis to identify cluster rules
○ Additional analysis such as association rules mining to identify
prominent rule patterns
○ Insights via a Clustering Dashboard
○ Tailored schemes of users
○ Graph Neural Network
○ Business metrics as defined by the end user need
○ Trigger retraining options
H2O.ai Confidential
sales@h2o.ai
Contact

More Related Content

Similar to [코세나, kosena] Auto ML, H2O.ai의 보험분야 활용사례

Data reply sneak peek: real time decision engines
Data reply sneak peek:  real time decision enginesData reply sneak peek:  real time decision engines
Data reply sneak peek: real time decision enginesconfluent
 
Cognitive Procurement Masterclass with IBM - SID 51774
Cognitive Procurement Masterclass with IBM - SID 51774Cognitive Procurement Masterclass with IBM - SID 51774
Cognitive Procurement Masterclass with IBM - SID 51774SAP Ariba
 
Warranty Predictive Analytics solution
Warranty Predictive Analytics solutionWarranty Predictive Analytics solution
Warranty Predictive Analytics solutionRevolution Analytics
 
Nurturing Digital Twins: How to Build Virtual Instances of Physical Assets to...
Nurturing Digital Twins: How to Build Virtual Instances of Physical Assets to...Nurturing Digital Twins: How to Build Virtual Instances of Physical Assets to...
Nurturing Digital Twins: How to Build Virtual Instances of Physical Assets to...Cognizant
 
Data analytics as a service
Data analytics as a serviceData analytics as a service
Data analytics as a serviceStanley Wang
 
DFS21_Main Stage_Steve Butcher_Microsoft_211130
DFS21_Main Stage_Steve Butcher_Microsoft_211130DFS21_Main Stage_Steve Butcher_Microsoft_211130
DFS21_Main Stage_Steve Butcher_Microsoft_211130FinTech Belgium
 
Data Robotics Brochure - Open Reply
Data Robotics Brochure - Open ReplyData Robotics Brochure - Open Reply
Data Robotics Brochure - Open ReplyAusrine S.
 
Big Data solution for multi-national Bank
Big Data solution for multi-national BankBig Data solution for multi-national Bank
Big Data solution for multi-national BankRitu Sarkar
 
Using Data Science to Build an End-to-End Recommendation System
Using Data Science to Build an End-to-End Recommendation SystemUsing Data Science to Build an End-to-End Recommendation System
Using Data Science to Build an End-to-End Recommendation SystemVMware Tanzu
 
Machine Learning in Customer Analytics
Machine Learning in Customer AnalyticsMachine Learning in Customer Analytics
Machine Learning in Customer AnalyticsCourse5i
 
Ai in insurance how to automate insurance claim processing with machine lear...
Ai in insurance  how to automate insurance claim processing with machine lear...Ai in insurance  how to automate insurance claim processing with machine lear...
Ai in insurance how to automate insurance claim processing with machine lear...Skyl.ai
 
Dataiku tatvic webinar presentation
Dataiku tatvic webinar presentationDataiku tatvic webinar presentation
Dataiku tatvic webinar presentationTatvic Analytics
 
Project Deliverable 4 Analytics, Interfaces, and Cloud Technolo.docx
Project Deliverable 4 Analytics, Interfaces, and Cloud Technolo.docxProject Deliverable 4 Analytics, Interfaces, and Cloud Technolo.docx
Project Deliverable 4 Analytics, Interfaces, and Cloud Technolo.docxwkyra78
 
Robotic process automation powers digital transformation in insurance industry
Robotic process automation powers digital transformation in insurance industryRobotic process automation powers digital transformation in insurance industry
Robotic process automation powers digital transformation in insurance industryArtivatic.ai
 
Sourcing & Procurement Analytics for the modern enterprise
Sourcing & Procurement Analytics for the modern enterpriseSourcing & Procurement Analytics for the modern enterprise
Sourcing & Procurement Analytics for the modern enterpriseBRIDGEi2i Analytics Solutions
 
How to Become an Analytics Ready Insurer - with Informatica and Hortonworks
How to Become an Analytics Ready Insurer - with Informatica and HortonworksHow to Become an Analytics Ready Insurer - with Informatica and Hortonworks
How to Become an Analytics Ready Insurer - with Informatica and HortonworksHortonworks
 
Insurance - Open Source Analytics Dashboards for Real Time Business Overview
Insurance - Open Source Analytics Dashboards for Real Time Business OverviewInsurance - Open Source Analytics Dashboards for Real Time Business Overview
Insurance - Open Source Analytics Dashboards for Real Time Business OverviewEuro IT Group
 
Intelligent underwriting workbench
Intelligent underwriting workbenchIntelligent underwriting workbench
Intelligent underwriting workbenchArtivatic.ai
 
DAC Tekiō by DAC Software Solutions Ltd.
DAC Tekiō by DAC Software Solutions Ltd.DAC Tekiō by DAC Software Solutions Ltd.
DAC Tekiō by DAC Software Solutions Ltd.Nicholai Portelli
 
Guardian analytics vs. actimize 2016
Guardian analytics vs. actimize 2016Guardian analytics vs. actimize 2016
Guardian analytics vs. actimize 2016Laurent Pacalin
 

Similar to [코세나, kosena] Auto ML, H2O.ai의 보험분야 활용사례 (20)

Data reply sneak peek: real time decision engines
Data reply sneak peek:  real time decision enginesData reply sneak peek:  real time decision engines
Data reply sneak peek: real time decision engines
 
Cognitive Procurement Masterclass with IBM - SID 51774
Cognitive Procurement Masterclass with IBM - SID 51774Cognitive Procurement Masterclass with IBM - SID 51774
Cognitive Procurement Masterclass with IBM - SID 51774
 
Warranty Predictive Analytics solution
Warranty Predictive Analytics solutionWarranty Predictive Analytics solution
Warranty Predictive Analytics solution
 
Nurturing Digital Twins: How to Build Virtual Instances of Physical Assets to...
Nurturing Digital Twins: How to Build Virtual Instances of Physical Assets to...Nurturing Digital Twins: How to Build Virtual Instances of Physical Assets to...
Nurturing Digital Twins: How to Build Virtual Instances of Physical Assets to...
 
Data analytics as a service
Data analytics as a serviceData analytics as a service
Data analytics as a service
 
DFS21_Main Stage_Steve Butcher_Microsoft_211130
DFS21_Main Stage_Steve Butcher_Microsoft_211130DFS21_Main Stage_Steve Butcher_Microsoft_211130
DFS21_Main Stage_Steve Butcher_Microsoft_211130
 
Data Robotics Brochure - Open Reply
Data Robotics Brochure - Open ReplyData Robotics Brochure - Open Reply
Data Robotics Brochure - Open Reply
 
Big Data solution for multi-national Bank
Big Data solution for multi-national BankBig Data solution for multi-national Bank
Big Data solution for multi-national Bank
 
Using Data Science to Build an End-to-End Recommendation System
Using Data Science to Build an End-to-End Recommendation SystemUsing Data Science to Build an End-to-End Recommendation System
Using Data Science to Build an End-to-End Recommendation System
 
Machine Learning in Customer Analytics
Machine Learning in Customer AnalyticsMachine Learning in Customer Analytics
Machine Learning in Customer Analytics
 
Ai in insurance how to automate insurance claim processing with machine lear...
Ai in insurance  how to automate insurance claim processing with machine lear...Ai in insurance  how to automate insurance claim processing with machine lear...
Ai in insurance how to automate insurance claim processing with machine lear...
 
Dataiku tatvic webinar presentation
Dataiku tatvic webinar presentationDataiku tatvic webinar presentation
Dataiku tatvic webinar presentation
 
Project Deliverable 4 Analytics, Interfaces, and Cloud Technolo.docx
Project Deliverable 4 Analytics, Interfaces, and Cloud Technolo.docxProject Deliverable 4 Analytics, Interfaces, and Cloud Technolo.docx
Project Deliverable 4 Analytics, Interfaces, and Cloud Technolo.docx
 
Robotic process automation powers digital transformation in insurance industry
Robotic process automation powers digital transformation in insurance industryRobotic process automation powers digital transformation in insurance industry
Robotic process automation powers digital transformation in insurance industry
 
Sourcing & Procurement Analytics for the modern enterprise
Sourcing & Procurement Analytics for the modern enterpriseSourcing & Procurement Analytics for the modern enterprise
Sourcing & Procurement Analytics for the modern enterprise
 
How to Become an Analytics Ready Insurer - with Informatica and Hortonworks
How to Become an Analytics Ready Insurer - with Informatica and HortonworksHow to Become an Analytics Ready Insurer - with Informatica and Hortonworks
How to Become an Analytics Ready Insurer - with Informatica and Hortonworks
 
Insurance - Open Source Analytics Dashboards for Real Time Business Overview
Insurance - Open Source Analytics Dashboards for Real Time Business OverviewInsurance - Open Source Analytics Dashboards for Real Time Business Overview
Insurance - Open Source Analytics Dashboards for Real Time Business Overview
 
Intelligent underwriting workbench
Intelligent underwriting workbenchIntelligent underwriting workbench
Intelligent underwriting workbench
 
DAC Tekiō by DAC Software Solutions Ltd.
DAC Tekiō by DAC Software Solutions Ltd.DAC Tekiō by DAC Software Solutions Ltd.
DAC Tekiō by DAC Software Solutions Ltd.
 
Guardian analytics vs. actimize 2016
Guardian analytics vs. actimize 2016Guardian analytics vs. actimize 2016
Guardian analytics vs. actimize 2016
 

More from kosena

빅재미(BigZAMi), 빅데이터 분석 플랫폼(AutoML포함) 소개해 드립니다.
빅재미(BigZAMi), 빅데이터 분석 플랫폼(AutoML포함) 소개해 드립니다.빅재미(BigZAMi), 빅데이터 분석 플랫폼(AutoML포함) 소개해 드립니다.
빅재미(BigZAMi), 빅데이터 분석 플랫폼(AutoML포함) 소개해 드립니다.kosena
 
[코세나, kosena] 생성AI 프로젝트와 사례
[코세나, kosena] 생성AI 프로젝트와 사례[코세나, kosena] 생성AI 프로젝트와 사례
[코세나, kosena] 생성AI 프로젝트와 사례kosena
 
[코세나, kosena] Auto ML, H2O.ai의 헬스케어분야 AI 활용 사례
[코세나, kosena] Auto ML, H2O.ai의 헬스케어분야 AI 활용 사례[코세나, kosena] Auto ML, H2O.ai의 헬스케어분야 AI 활용 사례
[코세나, kosena] Auto ML, H2O.ai의 헬스케어분야 AI 활용 사례kosena
 
[코세나, kosena] Auto ML, H2O.ai의 제조분야 AI 활용 사례
[코세나, kosena] Auto ML, H2O.ai의 제조분야 AI 활용 사례[코세나, kosena] Auto ML, H2O.ai의 제조분야 AI 활용 사례
[코세나, kosena] Auto ML, H2O.ai의 제조분야 AI 활용 사례kosena
 
[코세나, kosena] Auto ML 플랫폼 : H2O.ai DAI
[코세나, kosena] Auto ML 플랫폼 : H2O.ai DAI[코세나, kosena] Auto ML 플랫폼 : H2O.ai DAI
[코세나, kosena] Auto ML 플랫폼 : H2O.ai DAIkosena
 
[코세나, kosena] NFT Minting제언서
[코세나, kosena] NFT Minting제언서[코세나, kosena] NFT Minting제언서
[코세나, kosena] NFT Minting제언서kosena
 
[코세나, kosena] 산업부문별 인공지능 활용제안 가이드
[코세나, kosena] 산업부문별 인공지능 활용제안 가이드[코세나, kosena] 산업부문별 인공지능 활용제안 가이드
[코세나, kosena] 산업부문별 인공지능 활용제안 가이드kosena
 
[코세나, kosena] 금융권의 머신러닝 활용사례
[코세나, kosena] 금융권의 머신러닝 활용사례[코세나, kosena] 금융권의 머신러닝 활용사례
[코세나, kosena] 금융권의 머신러닝 활용사례kosena
 
[코세나, kosena] 빅데이터 구축 및 제안 가이드
[코세나, kosena] 빅데이터 구축 및 제안 가이드[코세나, kosena] 빅데이터 구축 및 제안 가이드
[코세나, kosena] 빅데이터 구축 및 제안 가이드kosena
 
[코세나, kosena] 빅데이터 기반의 End-to-End APM과 비정형 데이터 분석 자료입니다.
[코세나, kosena] 빅데이터 기반의 End-to-End APM과 비정형 데이터 분석 자료입니다.[코세나, kosena] 빅데이터 기반의 End-to-End APM과 비정형 데이터 분석 자료입니다.
[코세나, kosena] 빅데이터 기반의 End-to-End APM과 비정형 데이터 분석 자료입니다.kosena
 
[코세나, kosena] FDS(Fraud Detection System) Securities
[코세나, kosena] FDS(Fraud Detection System) Securities[코세나, kosena] FDS(Fraud Detection System) Securities
[코세나, kosena] FDS(Fraud Detection System) Securitieskosena
 

More from kosena (11)

빅재미(BigZAMi), 빅데이터 분석 플랫폼(AutoML포함) 소개해 드립니다.
빅재미(BigZAMi), 빅데이터 분석 플랫폼(AutoML포함) 소개해 드립니다.빅재미(BigZAMi), 빅데이터 분석 플랫폼(AutoML포함) 소개해 드립니다.
빅재미(BigZAMi), 빅데이터 분석 플랫폼(AutoML포함) 소개해 드립니다.
 
[코세나, kosena] 생성AI 프로젝트와 사례
[코세나, kosena] 생성AI 프로젝트와 사례[코세나, kosena] 생성AI 프로젝트와 사례
[코세나, kosena] 생성AI 프로젝트와 사례
 
[코세나, kosena] Auto ML, H2O.ai의 헬스케어분야 AI 활용 사례
[코세나, kosena] Auto ML, H2O.ai의 헬스케어분야 AI 활용 사례[코세나, kosena] Auto ML, H2O.ai의 헬스케어분야 AI 활용 사례
[코세나, kosena] Auto ML, H2O.ai의 헬스케어분야 AI 활용 사례
 
[코세나, kosena] Auto ML, H2O.ai의 제조분야 AI 활용 사례
[코세나, kosena] Auto ML, H2O.ai의 제조분야 AI 활용 사례[코세나, kosena] Auto ML, H2O.ai의 제조분야 AI 활용 사례
[코세나, kosena] Auto ML, H2O.ai의 제조분야 AI 활용 사례
 
[코세나, kosena] Auto ML 플랫폼 : H2O.ai DAI
[코세나, kosena] Auto ML 플랫폼 : H2O.ai DAI[코세나, kosena] Auto ML 플랫폼 : H2O.ai DAI
[코세나, kosena] Auto ML 플랫폼 : H2O.ai DAI
 
[코세나, kosena] NFT Minting제언서
[코세나, kosena] NFT Minting제언서[코세나, kosena] NFT Minting제언서
[코세나, kosena] NFT Minting제언서
 
[코세나, kosena] 산업부문별 인공지능 활용제안 가이드
[코세나, kosena] 산업부문별 인공지능 활용제안 가이드[코세나, kosena] 산업부문별 인공지능 활용제안 가이드
[코세나, kosena] 산업부문별 인공지능 활용제안 가이드
 
[코세나, kosena] 금융권의 머신러닝 활용사례
[코세나, kosena] 금융권의 머신러닝 활용사례[코세나, kosena] 금융권의 머신러닝 활용사례
[코세나, kosena] 금융권의 머신러닝 활용사례
 
[코세나, kosena] 빅데이터 구축 및 제안 가이드
[코세나, kosena] 빅데이터 구축 및 제안 가이드[코세나, kosena] 빅데이터 구축 및 제안 가이드
[코세나, kosena] 빅데이터 구축 및 제안 가이드
 
[코세나, kosena] 빅데이터 기반의 End-to-End APM과 비정형 데이터 분석 자료입니다.
[코세나, kosena] 빅데이터 기반의 End-to-End APM과 비정형 데이터 분석 자료입니다.[코세나, kosena] 빅데이터 기반의 End-to-End APM과 비정형 데이터 분석 자료입니다.
[코세나, kosena] 빅데이터 기반의 End-to-End APM과 비정형 데이터 분석 자료입니다.
 
[코세나, kosena] FDS(Fraud Detection System) Securities
[코세나, kosena] FDS(Fraud Detection System) Securities[코세나, kosena] FDS(Fraud Detection System) Securities
[코세나, kosena] FDS(Fraud Detection System) Securities
 

Recently uploaded

(Genuine) Escort Service Lucknow | Starting ₹,5K To @25k with A/C 🧑🏽‍❤️‍🧑🏻 89...
(Genuine) Escort Service Lucknow | Starting ₹,5K To @25k with A/C 🧑🏽‍❤️‍🧑🏻 89...(Genuine) Escort Service Lucknow | Starting ₹,5K To @25k with A/C 🧑🏽‍❤️‍🧑🏻 89...
(Genuine) Escort Service Lucknow | Starting ₹,5K To @25k with A/C 🧑🏽‍❤️‍🧑🏻 89...gurkirankumar98700
 
Adobe Marketo Engage Deep Dives: Using Webhooks to Transfer Data
Adobe Marketo Engage Deep Dives: Using Webhooks to Transfer DataAdobe Marketo Engage Deep Dives: Using Webhooks to Transfer Data
Adobe Marketo Engage Deep Dives: Using Webhooks to Transfer DataBradBedford3
 
Automate your Kamailio Test Calls - Kamailio World 2024
Automate your Kamailio Test Calls - Kamailio World 2024Automate your Kamailio Test Calls - Kamailio World 2024
Automate your Kamailio Test Calls - Kamailio World 2024Andreas Granig
 
Salesforce Certified Field Service Consultant
Salesforce Certified Field Service ConsultantSalesforce Certified Field Service Consultant
Salesforce Certified Field Service ConsultantAxelRicardoTrocheRiq
 
Russian Call Girls in Karol Bagh Aasnvi ➡️ 8264348440 💋📞 Independent Escort S...
Russian Call Girls in Karol Bagh Aasnvi ➡️ 8264348440 💋📞 Independent Escort S...Russian Call Girls in Karol Bagh Aasnvi ➡️ 8264348440 💋📞 Independent Escort S...
Russian Call Girls in Karol Bagh Aasnvi ➡️ 8264348440 💋📞 Independent Escort S...soniya singh
 
Professional Resume Template for Software Developers
Professional Resume Template for Software DevelopersProfessional Resume Template for Software Developers
Professional Resume Template for Software DevelopersVinodh Ram
 
Alluxio Monthly Webinar | Cloud-Native Model Training on Distributed Data
Alluxio Monthly Webinar | Cloud-Native Model Training on Distributed DataAlluxio Monthly Webinar | Cloud-Native Model Training on Distributed Data
Alluxio Monthly Webinar | Cloud-Native Model Training on Distributed DataAlluxio, Inc.
 
Der Spagat zwischen BIAS und FAIRNESS (2024)
Der Spagat zwischen BIAS und FAIRNESS (2024)Der Spagat zwischen BIAS und FAIRNESS (2024)
Der Spagat zwischen BIAS und FAIRNESS (2024)OPEN KNOWLEDGE GmbH
 
Try MyIntelliAccount Cloud Accounting Software As A Service Solution Risk Fre...
Try MyIntelliAccount Cloud Accounting Software As A Service Solution Risk Fre...Try MyIntelliAccount Cloud Accounting Software As A Service Solution Risk Fre...
Try MyIntelliAccount Cloud Accounting Software As A Service Solution Risk Fre...MyIntelliSource, Inc.
 
ODSC - Batch to Stream workshop - integration of Apache Spark, Cassandra, Pos...
ODSC - Batch to Stream workshop - integration of Apache Spark, Cassandra, Pos...ODSC - Batch to Stream workshop - integration of Apache Spark, Cassandra, Pos...
ODSC - Batch to Stream workshop - integration of Apache Spark, Cassandra, Pos...Christina Lin
 
办理学位证(UQ文凭证书)昆士兰大学毕业证成绩单原版一模一样
办理学位证(UQ文凭证书)昆士兰大学毕业证成绩单原版一模一样办理学位证(UQ文凭证书)昆士兰大学毕业证成绩单原版一模一样
办理学位证(UQ文凭证书)昆士兰大学毕业证成绩单原版一模一样umasea
 
chapter--4-software-project-planning.ppt
chapter--4-software-project-planning.pptchapter--4-software-project-planning.ppt
chapter--4-software-project-planning.pptkotipi9215
 
Folding Cheat Sheet #4 - fourth in a series
Folding Cheat Sheet #4 - fourth in a seriesFolding Cheat Sheet #4 - fourth in a series
Folding Cheat Sheet #4 - fourth in a seriesPhilip Schwarz
 
Asset Management Software - Infographic
Asset Management Software - InfographicAsset Management Software - Infographic
Asset Management Software - InfographicHr365.us smith
 
Building Real-Time Data Pipelines: Stream & Batch Processing workshop Slide
Building Real-Time Data Pipelines: Stream & Batch Processing workshop SlideBuilding Real-Time Data Pipelines: Stream & Batch Processing workshop Slide
Building Real-Time Data Pipelines: Stream & Batch Processing workshop SlideChristina Lin
 
Dealing with Cultural Dispersion — Stefano Lambiase — ICSE-SEIS 2024
Dealing with Cultural Dispersion — Stefano Lambiase — ICSE-SEIS 2024Dealing with Cultural Dispersion — Stefano Lambiase — ICSE-SEIS 2024
Dealing with Cultural Dispersion — Stefano Lambiase — ICSE-SEIS 2024StefanoLambiase
 
EY_Graph Database Powered Sustainability
EY_Graph Database Powered SustainabilityEY_Graph Database Powered Sustainability
EY_Graph Database Powered SustainabilityNeo4j
 
Unveiling Design Patterns: A Visual Guide with UML Diagrams
Unveiling Design Patterns: A Visual Guide with UML DiagramsUnveiling Design Patterns: A Visual Guide with UML Diagrams
Unveiling Design Patterns: A Visual Guide with UML DiagramsAhmed Mohamed
 
The Evolution of Karaoke From Analog to App.pdf
The Evolution of Karaoke From Analog to App.pdfThe Evolution of Karaoke From Analog to App.pdf
The Evolution of Karaoke From Analog to App.pdfPower Karaoke
 

Recently uploaded (20)

(Genuine) Escort Service Lucknow | Starting ₹,5K To @25k with A/C 🧑🏽‍❤️‍🧑🏻 89...
(Genuine) Escort Service Lucknow | Starting ₹,5K To @25k with A/C 🧑🏽‍❤️‍🧑🏻 89...(Genuine) Escort Service Lucknow | Starting ₹,5K To @25k with A/C 🧑🏽‍❤️‍🧑🏻 89...
(Genuine) Escort Service Lucknow | Starting ₹,5K To @25k with A/C 🧑🏽‍❤️‍🧑🏻 89...
 
Adobe Marketo Engage Deep Dives: Using Webhooks to Transfer Data
Adobe Marketo Engage Deep Dives: Using Webhooks to Transfer DataAdobe Marketo Engage Deep Dives: Using Webhooks to Transfer Data
Adobe Marketo Engage Deep Dives: Using Webhooks to Transfer Data
 
Automate your Kamailio Test Calls - Kamailio World 2024
Automate your Kamailio Test Calls - Kamailio World 2024Automate your Kamailio Test Calls - Kamailio World 2024
Automate your Kamailio Test Calls - Kamailio World 2024
 
Salesforce Certified Field Service Consultant
Salesforce Certified Field Service ConsultantSalesforce Certified Field Service Consultant
Salesforce Certified Field Service Consultant
 
Russian Call Girls in Karol Bagh Aasnvi ➡️ 8264348440 💋📞 Independent Escort S...
Russian Call Girls in Karol Bagh Aasnvi ➡️ 8264348440 💋📞 Independent Escort S...Russian Call Girls in Karol Bagh Aasnvi ➡️ 8264348440 💋📞 Independent Escort S...
Russian Call Girls in Karol Bagh Aasnvi ➡️ 8264348440 💋📞 Independent Escort S...
 
Professional Resume Template for Software Developers
Professional Resume Template for Software DevelopersProfessional Resume Template for Software Developers
Professional Resume Template for Software Developers
 
Alluxio Monthly Webinar | Cloud-Native Model Training on Distributed Data
Alluxio Monthly Webinar | Cloud-Native Model Training on Distributed DataAlluxio Monthly Webinar | Cloud-Native Model Training on Distributed Data
Alluxio Monthly Webinar | Cloud-Native Model Training on Distributed Data
 
Der Spagat zwischen BIAS und FAIRNESS (2024)
Der Spagat zwischen BIAS und FAIRNESS (2024)Der Spagat zwischen BIAS und FAIRNESS (2024)
Der Spagat zwischen BIAS und FAIRNESS (2024)
 
Try MyIntelliAccount Cloud Accounting Software As A Service Solution Risk Fre...
Try MyIntelliAccount Cloud Accounting Software As A Service Solution Risk Fre...Try MyIntelliAccount Cloud Accounting Software As A Service Solution Risk Fre...
Try MyIntelliAccount Cloud Accounting Software As A Service Solution Risk Fre...
 
ODSC - Batch to Stream workshop - integration of Apache Spark, Cassandra, Pos...
ODSC - Batch to Stream workshop - integration of Apache Spark, Cassandra, Pos...ODSC - Batch to Stream workshop - integration of Apache Spark, Cassandra, Pos...
ODSC - Batch to Stream workshop - integration of Apache Spark, Cassandra, Pos...
 
Hot Sexy call girls in Patel Nagar🔝 9953056974 🔝 escort Service
Hot Sexy call girls in Patel Nagar🔝 9953056974 🔝 escort ServiceHot Sexy call girls in Patel Nagar🔝 9953056974 🔝 escort Service
Hot Sexy call girls in Patel Nagar🔝 9953056974 🔝 escort Service
 
办理学位证(UQ文凭证书)昆士兰大学毕业证成绩单原版一模一样
办理学位证(UQ文凭证书)昆士兰大学毕业证成绩单原版一模一样办理学位证(UQ文凭证书)昆士兰大学毕业证成绩单原版一模一样
办理学位证(UQ文凭证书)昆士兰大学毕业证成绩单原版一模一样
 
chapter--4-software-project-planning.ppt
chapter--4-software-project-planning.pptchapter--4-software-project-planning.ppt
chapter--4-software-project-planning.ppt
 
Folding Cheat Sheet #4 - fourth in a series
Folding Cheat Sheet #4 - fourth in a seriesFolding Cheat Sheet #4 - fourth in a series
Folding Cheat Sheet #4 - fourth in a series
 
Asset Management Software - Infographic
Asset Management Software - InfographicAsset Management Software - Infographic
Asset Management Software - Infographic
 
Building Real-Time Data Pipelines: Stream & Batch Processing workshop Slide
Building Real-Time Data Pipelines: Stream & Batch Processing workshop SlideBuilding Real-Time Data Pipelines: Stream & Batch Processing workshop Slide
Building Real-Time Data Pipelines: Stream & Batch Processing workshop Slide
 
Dealing with Cultural Dispersion — Stefano Lambiase — ICSE-SEIS 2024
Dealing with Cultural Dispersion — Stefano Lambiase — ICSE-SEIS 2024Dealing with Cultural Dispersion — Stefano Lambiase — ICSE-SEIS 2024
Dealing with Cultural Dispersion — Stefano Lambiase — ICSE-SEIS 2024
 
EY_Graph Database Powered Sustainability
EY_Graph Database Powered SustainabilityEY_Graph Database Powered Sustainability
EY_Graph Database Powered Sustainability
 
Unveiling Design Patterns: A Visual Guide with UML Diagrams
Unveiling Design Patterns: A Visual Guide with UML DiagramsUnveiling Design Patterns: A Visual Guide with UML Diagrams
Unveiling Design Patterns: A Visual Guide with UML Diagrams
 
The Evolution of Karaoke From Analog to App.pdf
The Evolution of Karaoke From Analog to App.pdfThe Evolution of Karaoke From Analog to App.pdf
The Evolution of Karaoke From Analog to App.pdf
 

[코세나, kosena] Auto ML, H2O.ai의 보험분야 활용사례

  • 2. H2O.ai Confidential Democratizing AI for 20,000 Organizations 6 OF THE TOP INSURANCE Companies Partner with H2O.ai
  • 3. v H2O.ai Confidential Infrastructure and Operations Pricing & Underwriting Customer 360 Customer Behavioral Analysis Cyber-threat Detection Employee Retention Precision Pricing Automated underwriting Reserve Predictions Claims Management Call Centers/case routing Claims processing Damage Detection Estimation H2O.ai Use Case Catalog for Insurance Meeting the needs of Insurance Customers Fraud Claims Fraud Customer Experience KYC Automation Targeting Propensity to Buy Churn prediction Click on each for in depth coverage across Insurance use cases.
  • 5. v H2O.ai Confidential Cyber-threat detection Objective ● Help information security teams reduce risk and improve their security posture efficiently and effectively ● Automate threat detection across large numbers of devices, attack vectors, and data silos Outcome ● Continuous assessment of asset inventory to gain a complete and accurate view of devices, users and applications with access to IT systems ● Models that can detect and respond to deviations from the norm, even with noisy data ● Prediction models that will assess where and how a company is most likely to be breached, so planning and resource allocation can be directed toward weak points in the IT system ● Explainability of model recommendations and analysis for security operations leaders, CISOS, auditors and Board of Directors Business Value ● Get up to date knowledge of global and industry level threats to help prioritize defense systems ● Prevent cyber threat incidents and respond quicker/better when they do happen, improving OPEX ● Free up limited cybersecurity teams to focus on complicated cases, while AI takes care of routine tasks H2O.ai’s AI and Data Approaches ● Classification Models that can identify threatening vs non threatening events and actors ● Anomaly detection, entry classification, domain generation detection ● Unsupervised learning for unlabeled data, clustering data based on anomalies ● Analyze large data sets of events to identify many different types of threats (eg, malware, ransomware, email phishing, malicious code downloads) ● Train neural networks to tell the difference between malicious and safe files ● Use images to train classifier neural networks to detect malware in .doc and .pdf files ● Bias reduction models
  • 6. v H2O.ai Confidential Employee Retention Objective ● Predict the risk of employee departure and identify relevant factors ● View predictions of employee departure, forecast churn rates and identify relevant factors contained in employee data Outcome ● Models with high accuracy of predicting employee departure and identifying reasons for departure Business Value ● Retain flight-risk talent and thereby improve per employee efficiencies ● Minimize cost of employee replacement due to backfills H2O’s AI and Data Approaches ● Automatic feature engineering, model validation, model tuning, selection and deployment, machine learning interpretation (MLI), recipes, time series and auto pipeline generation for model scoring H2O AI App: Employee Churn Prediction
  • 8. v H2O.ai Confidential Precision Pricing Objective ● Increase efficiency and accuracy of the entire pricing process ● React faster to rate changes with less actuarial and IT effort ● Improve model explainability and transparency for all lines of business Outcome ● Virtually eliminate model bias with automatic disparate impact analysis for any model generated. ● Leverage new features gleaned from unstructured data through OCR and NLP ● Reduce IT resources required to deploy and monitor models with automated drift detection ● Automate the model documentation and filing process for state regulators Business Value ● Loss ratio improvement through enhanced segmentation and targeted pricing ● Assure regulators and public that with unsurpassed model explainability and interpretability ● Reduce risk and write more profitable, high quality business H2O.ai AI and Data Approaches ● Automate the preparation of data pulled from from multiple sources (eg, historical claims, policy and customer conversion and retention data), with custom recipes to build models with the insurance industry’s most widely used modeling platform ● Leverage automated feature engineering along with shapley values for enhanced model interpretability and accuracy ● Provide product managers with “what if” pricing simulation AI apps to predict impact of new pricing scenarios on profitability and retention
  • 9. v H2O.ai Confidential Automated Underwriting Objective ● Increase efficiency and accuracy of risk pricing process associated with traditional manual underwriting processes, increasing competitive advantage for insurers ● Widen the scope of data sources underwriters can use for evaluation Outcome ● Automate underwriting process including data collection, data extraction, and filling forms ● Extract information from unstructured data through OCR and NLP ● Make models that can learn from 3rd party data sources, claims histories and other past data to predict risk profile of new submissions ● Deeper visibility into customer risk profiles Business Value ● Tailor premiums to match each individual’s actual risk and optimize pricing ● Shorten underwriting workflows creating near real-time service for customer experience ● Improve underwriting profitability, while reducing OPEX, customer churn, and costs for customer retention H2O .ai AI and Data Approaches ● Use H2O recipes to automate the leveraging of data from multiple sources (historical policy data, policyholder churn & claims data to build models with LightGBM, Random Forest, XGBoost and neural networks that outperform traditional GLM approaches ● Train models with explainability and interpretability; understand variable importance ● Use deep learning to process images and raw machine data
  • 10. v H2O.ai Confidential Reserve Prediction Objective ● Use AI/Matching Learning to predict claim reserve amounts based on historical based trends ● Accurately estimate the Reserve, and recalibrating these estimates automatically when new information is added to the claim record helps carriers react swiftly to ever changing events Outcome ● Models that produce claims predictions that are aligned with actual claims on an individual claim level, both in expected value and variance. Business Value ● Ensure that the size of reservoir of money Insurance Carriers hold is adequately estimated the cost of future claims accurately, but not excessive, to cover potential liabilities already assumed ● Improved transparency on future liabilities in order to ensure proper statement of future liabilities on the balance sheet H2O .ai AI and Data Approaches ● Machine learning techniques on a single, end-to-end platform that leverages decision tree, SVM, neural network based and deep learning based techniques ● Joint modeling of several numerical values (eg, paid losses and claims outstanding) ● Make use of various claims data types (eg, numerical, categorical, textual data) that can enhance prediction model understanding and performance
  • 12. v H2O.ai Confidential Customer Handling and Call Center Optimization Objective ● Improve customer experience and reduce handling costs ● Map the query by the customer to appropriate resolution section Outcome ● An improvised automated method of Exploratory Data Analysis that segregates the data and provides numerical & graphical analysis based on the factors such as Date/Time, Location, Self Made Complaints, user profiles and past interactions. ● Improvised NLP techniques and real time processing can accurately pinpoint/classify the complaints into groups. ● Send high risk, complex complaints to the more experienced customer support agents. ● A multi-classification AI model to give suggestions for possible resolution. ● The model can also be used as a backend mapping engine for Chatbots and Customer Issue Resolution applications. ● The solution can be delivered through APIs and custom application Business Value ● Address the complaints of customers, maintain trust and ultimately - retain. ● Improve case deflection, the rate that customers are able to find their own answers to issues that they would have otherwise called support for. H2O’s AI and Data Approaches ● Conduct Exploratory Data Analysis on Customer Complaint Data. This data in the form of a Dataset will be Analysed, and numerical as well as graphical analysis will be provided. Answers to questions such as which location has the most complaints, what time period, is the complaint made on behalf of the user or made by the customer, etc. shall be provided in an organized manner. ● NLP Techniques will be used to process the data, which can be in the form of a dataset or realtime. ● H2O AI App, built with H2O Wave, an SDK for Data Scientists: ○ Import Data and Data Prep ( Complaint Data) ○ Exploratory Data Analysis to analyse the aforementioned points. ○ Interactive Dashboard where insights are displayed based on the searched tags, percentage rank of complaints ( #1 Rude behaviour, #2 Issues, etc.) There will also be percentage comparisons as the data increases ( Complaints in xxx domain are 15% higher, etc.) Top increases, top decreases ○ Natural Language Processing Apps ● Applying NLP models to obtain tags based on meanings. ○ Text Sentiment Apps
  • 13. v H2O.ai Confidential Claims Processing Objective ● Infuse the claims processing workflow using AI to inform adjusters with smart actions ● Quickly and accurately estimate and process claims ● Use AI to automate repetitive claims processes and streamline claims processing, allowing those with low risk to be processed automatically while higher risk claims are routed to investigators for review ● Predict claims payments at individual level Outcome ● An improvised automated method of Exploratory Data Analysis that segregates the data and provides numerical & graphical analysis based on factors such as Date/Time, Location, Self Made Complaints, user profiles and past interactions. ● An improvised automated method of Exploratory Data Analysis that segregates the data and provides numerical & graphical analysis based on factors such as Date/Time, Location, Self Made Complaints, user profiles and past interactions. ● NLP techniques and real time processing can accurately pinpoint/classify the complaints into groups. ● Send high risk, complex complaints to the more experienced customer support agents. ● A multi-classification AI model to give suggestions for possible resolution. ● The model can also be used as a backend mapping engine for Chatbots and Customer Issue Resolution applications. ● The solution can be delivered through APIs and custom application H2O’s AI and Data Approaches ● Build multiple models that can analyze all incoming structured and unstructured data, scanned images, faxes and natural language sets, policy information, medical bills reports, and legal documents to offer real time recommendations to claims adjusters. ● Each time new claims data is added to a record, apply algorithms to accurately score for fraud, reserve changes, and claim complexity at a level of nuance that is often missed through manual reviews. ● Build models that can route high complexity claims to seasoned adjuters for review, generating smart alerts with reason codes to take immediate action; Low complexity, low fraud claims are automatically flagged for straight through processing Business Value ● Fewer touches ● Claims adjusters react fast to critical events ● Faster claims settlement ● Reduced ALAE
  • 14. v H2O.ai Confidential Damage Detection and Estimation Objective ● Increase granularity and accuracy for damage estimation ● Accurately detect and classify damages from images Outcome ● Build models that can detect and classify damages from images, isolating where in the image there is damage and what type of damage is represented ● Translate damage into individual parts affected by adding additional information to the analysis, such as parts listed needed to repair, model optimum reconditioning scenarios ● Retrieve images of similar items (eg, vehicles), both damaged and undamaged, to help estimators evaluate whether or not the item was actually damaged Business Value ● Shorten processing time for property and casualty damage claims, improving customer satisfaction with faster, more accurate settlements ● Equip claims estimators with signals to identify where to focus attention during damage assessment, and reasons for it ● Route damage estimates to the right estimator team H2O’s AI and Data Approaches ● H2O’s image recognition and deep learning capabilities that can predict vehicle damage severity using multiclass classification ● Use binary classsification for vehicle damage identification ● Use multiclass classification of vehicle make, model, and year ● Use H2O Wave to make AI applications that allow business stakeholders to get: ○ a pre-selection of cases where pictures do not match with expectations ○ highlighted images where cases do not match claims reports ○ recommendations for next steps ● Explainable AI visualizations and narratives that explain why the model made its damage estimation
  • 15. Deliver on Customer Experience
  • 16. v H2O.ai Confidential Targeting Objective ● Send messages/content at the right time with the right content based on individual and collective customer behaviors ● Help marketing teams at FIs achieve more awareness, higher conversions and better engagements Outcome ● Pool large data sets to accurately predict custom audience segments with specific attributes relevant to the brand ● Identify unique and relevant audiences for the brand that will be in line with the audience data pools initially created Business Value ● Increase click through rates by 4x (industry expectations) ● Reduce costs per engagement H2O’s AI and Data Approaches ● Machine learning interpretability (MLI) that provides targeting explanations ● Ability to provide computational power to advance granular audience targeting capabilities ● Target identification models - automate and optimize processes for potential consumer identification, information extraction and market segmentation ● Probability models that will predict level at which consumers will click on ads, click through rates, etc ● Recommendation models - discover useful patterns to determine what the user finds interesting or not ● Machine learning interpretability that provides targeting explanations H2O AI Apps, built with H2O Wave, an SDK for Data Scientists: ● UMAP-based clustering App ● Customer Journey / Lead Scoring App
  • 17. v H2O.ai Confidential Objective ● Expand offers to customers with accuracy and speed by implementing predictive analytics ● Forecast behaviors of a target audience by analyzing their past behaviors Outcome ● Cut time to build models from 6 months to less than a week ● Doubled customer propensity to buy rate ● Realized additional revenue by being better able to target offers ● Created propensity models that helped companies identify the right customers and prospects that have a high likelihood to purchase a particular product or service Business Value ● Optimize promotions based on knowing which leads to sell to with the right message and product/services H2O’s AI and Data Approaches ● Create models and use custom recipes that generate features/variables that provide a probabilistic estimation of whether customers will perform any of such actions or not i.e. a propensity score. ● Use propensity scores to estimate value each customer brings in real-time. Make data available to Relationship Managers of the Bank (RMOs) who can further slice and dice the data and consume the information intuitively. ● H2O AI Apps, built with SDK for Data Scientists, can perform: ○ Data Preparation ○ Supervised Machine Learning ● Automatic feature engineering, model tuning and optimization, scoring pipeline generation ● Accurate time series capabilities ● Automatically generated visualizations and data plots ○ Nonlinear algorithmic modeling ○ Binary classification ○ Results viewed on Dashboards in a tabular format, grouped by different features (city, areacode, spending limit, etc) or for a single user; view target groups and selected features such as textual insights explaining the prediction, the propensity score for each user (mean or median) if a target group is selected. Propensity to Buy
  • 18. v H2O.ai Confidential Customer Churn Objective ● Predict customer churn based on behavioral patterns and cadence of transactions ● Identify leading indicators for churn, improving customer retention insights Outcome ● Optimized models up to the granularity of each customer ● Calculated daily probability for customer churn for early detection ● Distributed in-memory infrastructure trains and scores entire customer base in minutes instead of hours ● Reduced model building time from 6-7 hours to < 30 min Business Value ● Intercepted customer churn before it happened ● Know what leads a client towards the decision to leave the company. ● Develop loyalty programs and retention campaigns to keep as many customers as possible. ● Improving the customer retention rate for existing customers by just 5 percent can improve a company’s profitability by 25 to 95 percent (Bain & Co). H2O’s AI and Data Approaches ● Identify customers with high probability of churning by creating models and custom recipes specifically built for generating features/variables. ● Distributed algorithms (Random Forest, GBM) ● Capabilities to iterate through models with different parameters within time constraints, instead of relying on just one model ● H2O Customer Churn AI App, built with H2O Wave:
  • 20. v H2O.ai Confidential Claims Fraud Objective ● Find new ways to improve fraud detection accuracy and detection time across claims fraud types ● Use AI to generate new fraud rules ● Detect anomalous patterns for individuals, accounts and networks within and across portfolios ● Target different groups and subsets in a tailored approach, comparing different inferences from these subsets with the same inferences from the entire population ● Flag specific repetitive fraud groups and deploy counter measures Outcome ● 6x faster development of state-of-the-art ML models that pre-empted fraudulent claims H2O’s AI and Data Approaches ● Build AI models and use custom recipes specifically built for generating features/variables that provide associated information about fraudulent behavior. This data is then available to the fraud investigator who can further slice and dice the data and consume the information intuitively. ● Feature engineering with Deep Learning to model new and complex attack patterns quickly ● Behavior profiling for data networks - IP addresses, buying patterns ● Terabytes of data leveraged to deliver high scalability and performance, flexible deployment and integration with other big data frameworks ● Model Explainability aids investigators in understanding the why the model thinks there is potential fraud. Business Value ● Reduce fraud losses and improve customer experience ● 2x increase in accurate claims fraud detection ● 11% improvement in accuracy resulting in $1M saved monthly per basis point
  • 22. v H2O.ai Confidential Know your customer (KYC) Automation Objective ● Operationalize customer data to understand behavior analytics for several use cases across pricing, claims management, fraud prevention and more ● Identity verification Outcome ● Create a foundation for customer analytics ● sA series of models that can assess customer analytics, behaviors and needs across very different customer profiles Business Value ● Improve campaign effectiveness ● Accurate, targeted cross-sell / up-sell ● Retain your most profitable customers ● Deliver real-time superior customer experience at all points of service ● Increase customer lifetime value ● Verify that your customers are who they say they are and assessing the risks associated with each customer like the possibilities of fraud, money laundering, terrorism financing, etc. ● Improve digital onboarding experience with automated KYC processes, reducing application abandonment. H2O’s AI and Data Approaches ● Process large quantities of customer data and integrate into the H2O AI Cloud to build KYC models ● Probabilistic and Fuzzy Matching ● Document verification, NLP and image recognition to verify and authenticate customer information ● Seamlessly extract customer data and integrate with existing customer management or onboarding systems Customer Data Integration Social Profile Credit Profile Life Insurance Profile Mortgage Account Profile Bank Account Profile Customer 360º View Master Data Management H2O AI Engine / Probabilistic Matching
  • 23. v H2O.ai Confidential Customer Behavior Analysis Objective ● Segment users on the basis of different attributes such as - demographics data, transaction history, transaction behavior (Recency, Frequency, Monetary). ● Rapidly identify and evaluate customer behavior signals to take action on potential fraudulent activity, optimize marketing activities, etc Outcomes ● Create behavior scoring models ● Continuously update customer data Business Value ● Create foundation for: ○ Personalization strategies to improve customer experience, customer sentiment ○ Targeting different groups in a tailored approach ○ Recommending core banking products for cross-selling and up-selling, increase sales, revenues, and profits ○ Deploying customer churn prevention schemes H2O’s AI, Data-Driven Approach: ● Provide an end-to-end pipeline that collects and updates the customer data, performs data preparation, performs unsupervised machine learning, and generates clustering results. ● Use H2O Wave (SDK for Data Scientists) to build an AI App that can perform: ○ Data Preparation ○ Unsupervised Machine Learning ○ Rule fit analysis to identify cluster rules ○ Additional analysis such as association rules mining to identify prominent rule patterns ○ Insights via a Clustering Dashboard ○ Tailored schemes of users ○ Graph Neural Network ○ Business metrics as defined by the end user need ○ Trigger retraining options