Krishnan Jayaraman
Principal Consultant
Tech Mahindra
John Hardigree
Chief BDO
Revolution Analytics
Welcometo the session!
Experience the ease of an efficient AFTER SALES with
‘Warranty Predictive Analytics solution’
Moderator: Kannan Marghasahayam
Principal Consultant; TechMahindra
Agenda
Speaker Designation Agenda
Krishnan Jayaraman
Principal Consultant
Tech Mahindra
• Tech Mahindra’s lean Analytics Platform
• Warranty Analytics Solution Overview
John Hardigree Chief BDO
Revolution Analytics
• Revolution Analytics
• Warranty & Sensor Data Analysis Overview
Expert interaction – Q&A session
Lean Analytics- Simplify, Optimize,
Accelerate
 Simplify
• Common Storage Infrastructure
A Superior Way to Unlock Value from your Data - In a Box!
Business Insights
Business
Recommendations
Predictive Models
Analytics as a Service
Intellectual Property
• Analytics Platform
• Business Frameworks
• Data Management Platform
Shared Services
• Domain Experts
• Data Scientists
• Data Specialists
Storage Infrastructure
Authoring/Development Platform
 Optimize
• Shared Services
 Self Service
• Development Platform
 Accelerate
• Frameworks & IP
• People-centric model
• Data / Analytics-led approach
• Domain-specific or Project-specific methodology
• Capex. lead times and costs
• Pricing model based on Time and Material (T&M)
• 100% transfer of delivery/business risks to clients
Traditional
• Value-centric model
• IP and Consulting-led approach
• Cross-Domain and Best Practices methodology
• Opex. and business outcome based model
• Pricing model based on Value Realization and Proof of
Value Delivery (POVs).
• Minimal transfer of delivery/business risks
Lean
Lean is Better !
Accelerated Value Delivery
More than 50%reductionin time to delivery analytics services
Traditional Analytics Delivery Approach
Engage with
Business for
Hypothesis
Define the
KPIs from
multiple client
meetings
Validate the
KPIs and
finalize the
list
Seek data
and transform
as per KPIs
Model creation w ith
KPIs and refine
based on business
understanding
Show the output of
model and seek
business feedback
Business value
realized post
solution
deployment
TechM’s Lean Analytics Approach
Engage w ith business
to tick the predefined
KPIs from the
framew ork
Seek the
available data
points from client
Model creation and review
the output simultaneously
w ith the business for
validation and refinement
Deliver reports
and
dashboards
Business value
delivered immediately
after business GO
Our Value proposition
Lean Analytics can significantly reduce the time to delivery and
Business Risk
Detail Reference Architecture
Key benefits of Lean Analytics to Business Value Delivery
Accelerate Analytics Value
Potential acceleration through
analytics framework
standardization
Head start with built in business
KPIs Framework and Industry
Domain Use Cases
Reduced time to market by 50%
Optimize workload for Data
Scientists
Reuse Analytics Assets
Reuse existing knowledge base (Assets
– KPIs) for newuse case creation
Reuse existing use cases for faster time
to delivery and optimization
Leverage internal asset base created
within the organization over a period of
time
Single point Analytical Model
Repository
Analytics Governance
Single point Analytics KPI repository
End-to-End business users’ control on
analytics delivery
Traceability of legacy use case
deployment
Standardization of predictive analytical
model building initiatives
Focus on core skills: Business Insights
and Analysis
Lean Analytics: Simplify, Optimize,Accelerate Analytics Value Delivery
Tech Mahindra- Lean Analytics Platform
Tech Mahindra - Lean Analytics Platform
Business Solution
 Proactively identifies systematic
error patterns and their
dependencies, hence enabling to
reduce failure.
 Enabling the finance department to
estimate the future cost of
warranty claims, so to proactively
arrange warranty reserve fund.
 Helps to identify optimal warranty
term based on cost, improving
terms and conditions for supplier,
manufacturer and customers.
 Helps organizations to plan their
inventory of spare parts & labor
required accordingly.
 The claim rates of each month are taken as time
series data, considering Trend, Seasonality,
Cyclicity and some irregularity, data also includes
parameters like labor hours spent, parts used,
operations performed
 Modeling Technique : Multiple forecasting models
have been built using Arimax (Auto Regressive
Integrated Moving Average with External Variables)
& linear Regression
 Solution not only predicts the future claim cost but
also provide multiple forecast at various dimensions
like the number of vehicles to come for claim in
future in different vehicle categories, under different
claim types
Problem Statement Business BenefitsSolution Developed
Warranty Forecasting Analytics
Data Preparation
Seasonality and Cyclicality Analysis
Predictive Analysis
Time Series Data Analytical Engine Final Outcome analysis & dashboards
Forecasting
 Warranty claim cost are becoming one
of the major factor of concern for the
automobile companies due to higher
cost involved & increasingly more
incidents of consumer coming for
claims under warranty.
 The automobile manufacturers need a
predictive analytics platform to know
more about future scenarios about the
claims cost so as to plan their fiscal
year budgets & take pro-active actions
accordingly.
 The solution predicts the warranty cost
for different dimensions like warranty
cost for a region & vehicle models.
12
OUR COMPANY
The leading provider
of advanced analytics
software and services
based on open source R,
since 2007
OUR PRODUCT
REVOLUTION R: The
enterprise-grade predictive
analytics application platform
based on the R language
SOME KUDOS
Visionary
Gartner Magic Quadrant
for Advanced Analytics
Platforms, 2014
13
Game Changing Analytics Examples
 Marketing: Clickstream& Campaign Analyses
 Digital Media: Recommendation Engines
 Social Media: Sentiment Analysis
 Retail: Purchase Prediction
 Insurance: Fraud Waste and Abuse
 Healthcare Delivery: Treatment Outcome Prediction
 Risk Analysis: Insurance Underwriting
 Manufacturing: Predictive Maintenance
 Operations: Supply Chain Optimization
 Econometrics: Market Prediction
 Marketing: Mix and Price Optimization
 Life Sciences: Pharmacogenetics
 Transportation: Asset Utilization
Model Development for Vehicle Data Analysis
Profile: TheAnalytics R&D team of the multinationalautomobile
manufacturerworked with RevolutionAnalytics Consultants to
perform Survival Analysis, and to buildand deploy DecisionTrees
and TimeSeriesmodels
Key Technologyand Services:Revolution R Enterprise for
Big Data Analytics, Consulting, Training
AnalyticApproach –WarrantyDataAnalysis: Estimating
the life of an automobilecomponentusingSurvival Analysis
with Cox proportionalhazards. Modelsare trained using
historicaldata, consistingof warranty claims,and regionand
weather related variables suchsnow, rain, temperature etc.
Outcome: New analytics paradigm for existing processes
introduced,with potential for millionsof dollars in cost
savings through improved warranty contracts, and re-
designedautomobilecomponents.
>Warranty & Sensor Data Analysis
>R/Revolution R Enterprise Training
AnalyticApproach –SensorDataAnalysis: Use sensor data
from vehicle componentsto build DecisionTreesfor classification,
and to establish range of predictedvalues for sensor readingsso
that actualreadings canbe analyzed for outliers.
Bottomline: New analytics initiative for buildingan intelligent
automobilesystem that’s capableof guidingthe driver upon
detectionof anomaliesin driving patterns.
“Theconsultantsand training instructorsfrom Revolution
Analyticswere very knowledgeableand supported mevery
well.I am lookingforward to taking mylearningsto the larger
analyticsteam at mycompany.” SeniorResearcher,Analytics
R&D
Lean Analytics Platform
Sample Screen Shots
Potential Impacts
Improved warranty cost accuracy
Potential Impacts
RCA analysis for significant contributors
Potential Impacts
Reduction in significant warranty cost.
Potential Impacts
Defect patterns and dependencies.
Potential Impacts
Reduce failures and improve customer satisfaction.
Potential Impacts
Improve process efficiency.
Q & A
session
For Demo, Please send request to:
• Pitchcommunications@techmahindra.com
• Johnhard@microsoft.com
• Krishnan.Jayaraman@TechMahindra.com

Warranty Predictive Analytics solution

  • 1.
    Krishnan Jayaraman Principal Consultant TechMahindra John Hardigree Chief BDO Revolution Analytics Welcometo the session! Experience the ease of an efficient AFTER SALES with ‘Warranty Predictive Analytics solution’ Moderator: Kannan Marghasahayam Principal Consultant; TechMahindra
  • 2.
    Agenda Speaker Designation Agenda KrishnanJayaraman Principal Consultant Tech Mahindra • Tech Mahindra’s lean Analytics Platform • Warranty Analytics Solution Overview John Hardigree Chief BDO Revolution Analytics • Revolution Analytics • Warranty & Sensor Data Analysis Overview Expert interaction – Q&A session
  • 3.
    Lean Analytics- Simplify,Optimize, Accelerate  Simplify • Common Storage Infrastructure A Superior Way to Unlock Value from your Data - In a Box! Business Insights Business Recommendations Predictive Models Analytics as a Service Intellectual Property • Analytics Platform • Business Frameworks • Data Management Platform Shared Services • Domain Experts • Data Scientists • Data Specialists Storage Infrastructure Authoring/Development Platform  Optimize • Shared Services  Self Service • Development Platform  Accelerate • Frameworks & IP
  • 4.
    • People-centric model •Data / Analytics-led approach • Domain-specific or Project-specific methodology • Capex. lead times and costs • Pricing model based on Time and Material (T&M) • 100% transfer of delivery/business risks to clients Traditional • Value-centric model • IP and Consulting-led approach • Cross-Domain and Best Practices methodology • Opex. and business outcome based model • Pricing model based on Value Realization and Proof of Value Delivery (POVs). • Minimal transfer of delivery/business risks Lean Lean is Better !
  • 5.
    Accelerated Value Delivery Morethan 50%reductionin time to delivery analytics services Traditional Analytics Delivery Approach Engage with Business for Hypothesis Define the KPIs from multiple client meetings Validate the KPIs and finalize the list Seek data and transform as per KPIs Model creation w ith KPIs and refine based on business understanding Show the output of model and seek business feedback Business value realized post solution deployment TechM’s Lean Analytics Approach Engage w ith business to tick the predefined KPIs from the framew ork Seek the available data points from client Model creation and review the output simultaneously w ith the business for validation and refinement Deliver reports and dashboards Business value delivered immediately after business GO Our Value proposition Lean Analytics can significantly reduce the time to delivery and Business Risk
  • 6.
  • 7.
    Key benefits ofLean Analytics to Business Value Delivery Accelerate Analytics Value Potential acceleration through analytics framework standardization Head start with built in business KPIs Framework and Industry Domain Use Cases Reduced time to market by 50% Optimize workload for Data Scientists Reuse Analytics Assets Reuse existing knowledge base (Assets – KPIs) for newuse case creation Reuse existing use cases for faster time to delivery and optimization Leverage internal asset base created within the organization over a period of time Single point Analytical Model Repository Analytics Governance Single point Analytics KPI repository End-to-End business users’ control on analytics delivery Traceability of legacy use case deployment Standardization of predictive analytical model building initiatives Focus on core skills: Business Insights and Analysis Lean Analytics: Simplify, Optimize,Accelerate Analytics Value Delivery
  • 8.
    Tech Mahindra- LeanAnalytics Platform
  • 9.
    Tech Mahindra -Lean Analytics Platform
  • 10.
    Business Solution  Proactivelyidentifies systematic error patterns and their dependencies, hence enabling to reduce failure.  Enabling the finance department to estimate the future cost of warranty claims, so to proactively arrange warranty reserve fund.  Helps to identify optimal warranty term based on cost, improving terms and conditions for supplier, manufacturer and customers.  Helps organizations to plan their inventory of spare parts & labor required accordingly.  The claim rates of each month are taken as time series data, considering Trend, Seasonality, Cyclicity and some irregularity, data also includes parameters like labor hours spent, parts used, operations performed  Modeling Technique : Multiple forecasting models have been built using Arimax (Auto Regressive Integrated Moving Average with External Variables) & linear Regression  Solution not only predicts the future claim cost but also provide multiple forecast at various dimensions like the number of vehicles to come for claim in future in different vehicle categories, under different claim types Problem Statement Business BenefitsSolution Developed Warranty Forecasting Analytics Data Preparation Seasonality and Cyclicality Analysis Predictive Analysis Time Series Data Analytical Engine Final Outcome analysis & dashboards Forecasting  Warranty claim cost are becoming one of the major factor of concern for the automobile companies due to higher cost involved & increasingly more incidents of consumer coming for claims under warranty.  The automobile manufacturers need a predictive analytics platform to know more about future scenarios about the claims cost so as to plan their fiscal year budgets & take pro-active actions accordingly.  The solution predicts the warranty cost for different dimensions like warranty cost for a region & vehicle models.
  • 12.
    12 OUR COMPANY The leadingprovider of advanced analytics software and services based on open source R, since 2007 OUR PRODUCT REVOLUTION R: The enterprise-grade predictive analytics application platform based on the R language SOME KUDOS Visionary Gartner Magic Quadrant for Advanced Analytics Platforms, 2014
  • 13.
    13 Game Changing AnalyticsExamples  Marketing: Clickstream& Campaign Analyses  Digital Media: Recommendation Engines  Social Media: Sentiment Analysis  Retail: Purchase Prediction  Insurance: Fraud Waste and Abuse  Healthcare Delivery: Treatment Outcome Prediction  Risk Analysis: Insurance Underwriting  Manufacturing: Predictive Maintenance  Operations: Supply Chain Optimization  Econometrics: Market Prediction  Marketing: Mix and Price Optimization  Life Sciences: Pharmacogenetics  Transportation: Asset Utilization
  • 14.
    Model Development forVehicle Data Analysis Profile: TheAnalytics R&D team of the multinationalautomobile manufacturerworked with RevolutionAnalytics Consultants to perform Survival Analysis, and to buildand deploy DecisionTrees and TimeSeriesmodels Key Technologyand Services:Revolution R Enterprise for Big Data Analytics, Consulting, Training AnalyticApproach –WarrantyDataAnalysis: Estimating the life of an automobilecomponentusingSurvival Analysis with Cox proportionalhazards. Modelsare trained using historicaldata, consistingof warranty claims,and regionand weather related variables suchsnow, rain, temperature etc. Outcome: New analytics paradigm for existing processes introduced,with potential for millionsof dollars in cost savings through improved warranty contracts, and re- designedautomobilecomponents. >Warranty & Sensor Data Analysis >R/Revolution R Enterprise Training AnalyticApproach –SensorDataAnalysis: Use sensor data from vehicle componentsto build DecisionTreesfor classification, and to establish range of predictedvalues for sensor readingsso that actualreadings canbe analyzed for outliers. Bottomline: New analytics initiative for buildingan intelligent automobilesystem that’s capableof guidingthe driver upon detectionof anomaliesin driving patterns. “Theconsultantsand training instructorsfrom Revolution Analyticswere very knowledgeableand supported mevery well.I am lookingforward to taking mylearningsto the larger analyticsteam at mycompany.” SeniorResearcher,Analytics R&D
  • 15.
  • 16.
  • 17.
    Potential Impacts RCA analysisfor significant contributors
  • 18.
    Potential Impacts Reduction insignificant warranty cost.
  • 19.
  • 20.
    Potential Impacts Reduce failuresand improve customer satisfaction.
  • 21.
  • 22.
    Q & A session ForDemo, Please send request to: • Pitchcommunications@techmahindra.com • Johnhard@microsoft.com • Krishnan.Jayaraman@TechMahindra.com