using technology and big data to provide our customers with a … · 2017. 9. 21. · 19-09-2017...
TRANSCRIPT
Using Technology and Big Data to Provide our
Customers with a Passenger Experience
Presented by Suresh Kakarla
CEO of TollPlus LLC
19-09-2017 TollPlus LLC 2017 1
Analytics: Current Industry Context
9/19/2017 TollPlus LLC. 2
3
4
2
Weak Performers Strong Performers
Analytics: Current Issues
9/19/2017 TollPlus LLC. 3
Limited Data Sources
Mostly Manual Analysis Effort
Low Confidence Recommendations
Extremely Difficult to Predict Customer Behavior
Long Durations for Analysis and Research
Does Not Consider Variability in Research Parameters
Impossible to Determine Cluster Classifications To Determine Correlated Usage Patterns
Advanced Analytics Model
9/19/2017 TollPlus LLC. 4
Analyze Business Needs
•Analyze Current Processes
•Identify Key Needs
Collect And Prepare Data
•Identify Data Sources
•Validate And Clean Data
Perform Exploratory Analysis
•Data Distribution Analysis
•Eliminate Outliers
Advanced Analytics Engine
•Clustering / Classifications
•Correlations
•Predictive/Forecasting
•Recommendations
Provide Insights
•Recommendations/Best Actions
•Business Strategy Advice
Internal
External
Graphical &
Desktop tools
Domain
Expertise
Data
Visualization Data Sciences
Process Framework
Multiple Regression
Logical Regression
Neural Networks
Decision Trees
1
2
34
5
Integration Engine
Customer Mobility and
Big Data Ecosystem
19-09-2017 TollPlus LLC 2017 5
Scoop
(Relational
/ CSV)
Flume /
Kafka
(Streams)
Custom
Scripts
Data
Ingest &
Process
Map
Reduce
Engine
Spark
Engine
Subjective
Data Mart
Subjective
Data Mart
Mongo
DB
MySQL
Elastic
Hadoop
FS
HBase
Staging / Collection layerTraffic
Rate
Demographics
Customer Feedback
Billing
Interoperability
Payments
Violations
Tag
Cost
DMV
DOT
Address Verification
Email verification
Domain Model Data
Analysis Specific Data
High Quality Data for Processing
Summarized/Aggregated/
Composed Data
Meta Data for Processing and
Tools
Indexed and Sharded Data
Reference Data
Event Data
Tool for ingesting
relational & character
separated files data
Tool for processing
streaming and event
information
Python, shell scripts or
programs to cleanse
incoming data and implement
security, file transfer, etc
Integration Engine
Customer Mobility and
Big Data Ecosystem
19-09-2017 TollPlus LLC 2017 6
Scoop
(Relational
/ CSV)
Flume /
Kafka
(Streams)
Custom
Scripts
Data
Ingest &
Process
Map
Reduce
Engine
Spark
Engine
Subjective
Data Mart
Subjective
Data Mart
Mongo
DB
MySQL
Elastic
Hadoop
FS
HBase
Staging / Collection layerTraffic
Rate
Demographics
Customer Feedback
Billing
Interoperability
Payments
Violations
Tag
Cost
DMV
DOT
Address Verification
Email verification
Domain Model Data
Analysis Specific Data
High Quality Data for Processing
Summarized/Aggregated/
Composed Data
Meta Data for Processing and
Tools
Indexed and Sharded Data
Reference Data
Event Data
Document database for
staging & storing summary
information
Index database, for best search response and
reference data
Store all data, including unformatted data,
warehouse @ lowest cost
Provide definition to Hadoop data, enable queries with SQL like
syntax
Integration Engine
Customer Mobility and
Big Data Ecosystem
19-09-2017 TollPlus LLC 2017 7
Scoop
(Relational
/ CSV)
Flume /
Kafka
(Streams)
Custom
Scripts
Data
Ingest &
Process
Map
Reduce
Engine
Spark
Engine
Subjective
Data Mart
Subjective
Data Mart
Mongo
DB
MySQL
Elastic
Hadoop
FS
HBase
Staging / Collection layerTraffic
Rate
Demographics
Customer Feedback
Billing
Interoperability
Payments
Violations
Tag
Cost
DMV
DOT
Address Verification
Email verification
Domain Model Data
Analysis Specific Data
High Quality Data for Processing
Summarized/Aggregated/
Composed Data
Meta Data for Processing and
Tools
Indexed and Sharded Data
Reference Data
Event Data
Lowest latency, in memory engine for
transformations & processing
The datamarts are created to store summary & final
information on i.e. customer notifications to
be sent today, relevant history, etc
Implement map reduce algo to find
answers
Statistical Intelligence by Toll Agency
Customer Mobility and
Big Data Ecosystem (Cont’d..)
19-09-2017 TollPlus LLC 2017 8
Mongo
DB
MySQL
Elastic
Hadoop
FS
HBase
Map Reduce
Engine
Spark
Engine
Subjective
Data Mart
Subjective
Data Mart
R
Analysis
Visualizations
Outcomes
Statistical Analysis
Domain Model Data
Analysis Specific Data
High Quality Data for
Processing
Summarized/Aggregated
/Composed Data
Meta Data for Processing
and Tools
Indexed and Sharded
Data
Reference Data
Event Data
Revenue Projection Intelligence
Pricing Efficiency
Customer Value Add Services
Customer Mobility Information
Services for Retail Sector
Vehicle Movement Intelligence
Services for Govt Sector
Vehicle Pattern Notification
Services for Logistics Sector
Kibana, X-pack and related tools
This is a bevy of statistical model tools that would do
predications with their occurance probability,
comparirion anlysis, etc
This is the tool to enable statistical analysis
ecosytem
Revenue Projection
19-09-2017 TollPlus LLC 2017 9
Financial Prediction
Models
Statistical Analysis
TechniquesProbability
• Revenue Projections
• What if analysis
• Exceptions alert
• Comparative analysisBenefits for Toll Agency• Forecast revenue based on factors that are non-
financial and qualitative
• Simulate the impact of adding a lane on revenue
• Conduct what if analysis and comparisons
• Ability to measure the accuracy of the metrics
calculated
1Revenue Details (Historical)
2Cost Details (Historical)
3Other Financial Data
4Traffic Data
5DMV Information
6Other Market Data
This is a multi variable analysis, in which financial bond performance data will be modified based on non financial items in
isolation or in conjunction. They will define actions for the toll agency and help in making decisions to achieve any dynamic goals. This also can be used to measure the performance of actions taken
and their effectiveness within a time slice. This can identify/indicate if any factors that have not been considered or
misinterpreted in the causal analysis that is done in the past
Customer Value Added Services
19-09-2017 TollPlus LLC 2017 10
Decision Tree
processing
Renewal
Notification to
Customer
By Toll Agency
Auto Renewal by
Agency
Benefits for Customer• Improved Customer Experience
• Reduction in Travel Times
• Auto renewal/replenishment
Benefits for Toll Agency• Revenue & Traffic Forecasting
• Additional Revenues
• Reduced Congestion
• Reduction in Operational Costs
Customer Service
Request for
Auto Renewal through
Toll Agency
DMV
1 Demographics
2 Billing
3 Violations
4 Tag
5 DMV/DOT
6 USPS
7 Transit
8 Address Verification
9 Email Verification
Customer lifestyle improvements
19-09-2017 TollPlus LLC 2017 11
Notification to
Customer
By Toll Agency
Benefits for Customer• Notification 4 to 6 Hours in-advance, for traffic pattern
change, accidents, etc.
• Personalized travel time based information
• Better Information to manage life better
Benefits for Toll Agency• Better revenue from uniform traffic patterns
• Better congestion management
• Better customer service
1 Demographics
2 Billing
3 Violations
4 Tag
5 Market data
6 Public Event information
7 Transit
8 Geo positioning / maps data
Predictive Analysis
Event processing
Exception determination
Customer Mobility Information services
19-09-2017 TollPlus LLC 2017 12
Benefits for Retail Company• Effective Campaign management and
Promotions for road customers
• Use Customized & Intelligent Billboards
• Better Anticipate Customer Volumes
Associative Analyzed
Cluster Analyzed
Retail Company
Benefits for the Toll Agency• Additional Revenue
1 Traffic
2 Demographics
3 Rate
4 Customer Feedback
5 Billing
6 Interoperability
7 Payments
8 Violations
9 Tag
10 Maps/Geo Positioning
Knowledge of customer segmentation with respect to time can generate complex campaigns, which will help retail industry and the
customer both. E.g. Big events like a particular rock concert is the only place where a customer category adverts occur, here they will
know which day and which time and where a segment of customers are by using the agency data. The customer is completely
deidentified and personal information is scrubbed in compliance with govt regulations.
Vehicle Pattern Notification Services
19-09-2017 TollPlus LLC 2017 13
Processed & Segmented Data/
Streams
Logistics
Benefit for the Logistics Company• Optimized Road Vehicles
Operational Efficiency
• Optimized Cost Management
• Determination of Operation Actions
• Discovery of Opportunities
Benefit for the Toll Agency• Additional Revenue
1 Traffic
2 Demographics
3 Rate
4 Customer Feedback
5 Billing
6 Interoperability
7 Payments
8 Violations
9 Tag
10 Maps/Geo Positioning
Vehicle Movement
Intelligence Services
19-09-2017 TollPlus LLC 2017 14
Notification
engine
Alert
Prediction Engine
Benefits for Customer• Alerted on Traffic congestions
• Alert on Rate changes
• Alert on Emergency Conditions
• Improved Customer Experience
Benefits for Toll agency• Better Traffic Management
• Increased Revenue
• Increased Customer Subscription
Benefits for Govt Sector• Alerts to Fire Department on
Hazardous Vehicles
• Alerts Police Department for
Criminal Tracking
• Alert on Emergency Conditions
1 Rate
2 Traffic
3 DMV/DOT
4 Weather Data
5 Events Information
6 Market Analysis Data
7 Other Market Data
8 Analysis Summary Data at Toll Agency
Bigdata Program
Roadmap for a Toll Agency
19-09-2017 TollPlus LLC 2017 15
Budgeting and Cost Benefit Analysis
Business Case Specification
Corporate Data Dictionary
Specification
Organization User Identification
Identification for Operational
Procedures for Analytics
Configure Business Data
Analytics Specifications
Configure Clustered Software
System
Provision Hardware and
Software for the Analytics needs
Identification and Integration of
Organizational Data Across Various
Sources for Analytics
Define and Implement
Operational Processes for
Hardware Usage
Define and Implement
Monitoring and Error Resolution
Tools and Processes
Quick Refinement of Turn-around
Cycle Management Processes
Define Change Management,
Operational Practices
Strategic Implementation Operational
Bigdata Challenges
for a Toll Agency
19-09-2017 TollPlus LLC 2017 16
Strategic
Implementation
Operational
Bigdata Cost Benefit Analysis: A Forecasting Challenge
Alignment with Organization Strategy Needs Significant Thinking
Stakeholders Identification and Education will Require Good Planning and Coordination
Pooling Data Sources Require Significant Effort and Turn Around
Usage of Complex Statistical Methods and Tools will Need Higher Expertise
Legacy Systems: Integration and Feasibility Issues
Usage of Cutting Edge Technology Tools will Lead to Feasibility and Stability Issues
Monitoring Challenges will increase the Operational Burden
Error Recovery will involve significant Effort
Exponential Increase in Analysis Needs over Short Periods of Time will Impact Overall Timelines
Due to Significant Insight into Business from a Basket of New Perspectives and Questions will be Difficult to Manage