TMW Systems (a Trimble Company) has been in the business of long-haul trucking, logistics operations and fleet management for more than thirty years, but we wanted more data, so we turned to our customer community. Now, we turn that data into market intelligence, which we then provide back to our customers. To do this, we invested heavily in Hortonworks Data Platform running on Microsoft Azure in the cloud. In our talk, we’ll share our strategy for capturing operational, maintenance, financial and mobile communications information and how we provide that back to our customer base. Our approach enables advanced analytics by leveraging Big Data technologies to find new relationships in data that may have been previously overlooked. Survey responses capture business performance metrics, strategy and emerging trends from 150 businesses, representing more than 31 billion dollars in freight movement. Learn how we combine that survey data with other sources like machine and sensor data to help guide our customers to profitability.

1. How Market Intelligence From Hadoop on Azure
Shows Trucking Companies a Clear Road to
Profitability
Timothy Leonard, EVP
Dr. Piyush Kumar, Data Scientist
JUNE 2017

2. Industry/Business Problem

3. Marketing intelligence is turning data into insights and opportunities
MARKET
INTELLIGENCE
TRANSPORTATION
DATA
USER
DATA
SERVICE
DATA
EXTERNAL
DATA
Events
Lanes
Rates
Drivers
Payment terms
Shipments
Planning Time
Reports
Response Time
On-Time Pickup
On-Time Del.
Payment Terms
Network
Telematics
CSA
Social Media
Marketing
campaign
Strategic
planning
Data-driven
marketing
Predictive &
prescriptive analytics
Effective resource
planning
Increase customer
retention
Lead scoring
Visibility to industry
benchmarks
Increase upselling
& cross-selling
Opportunities

4. It is not simple….involves analyzing petabytes of data (~2 trillion records)
Industry TL & LTL & PF Home delivery TL & LTL & PF Home delivery TL & LTL & PF Home Delivery
Market Penetration (worst-case) 100% 5%
Drivers 3,500,000 100,000,000 3,500,000 100,000,000 3,500,000 100,000,000
Assets 2,700,000 235,000,000 2,700,000 235,000,000 2,700,000 235,000,000
Loads 2,700,000 5,000,000 243,000,000 450,000,000 1,080,000,000 2,000,000,000
Orders (Loads x 3) 8,100,000 10,000,000 729,000,000 900,000,000 3,240,000,000 4,000,000,000
Commodity (Orders x 3) 24,300,000 30,000,000 2,187,000,000 2,700,000,000 9,720,000,000 12,000,000,000
User Fields (Orders x 5) 40,500,000 50,000,000 3,645,000,000 4,500,000,000 16,200,000,000 20,000,000,000
Order Audit (Orders x 20) 162,000,000 200,000,000 14,580,000,000 18,000,000,000 64,800,000,000 80,000,000,000
Records (Orders x 3) 24,300,000 30,000,000 2,187,000,000 2,700,000,000 9,720,000,000 12,000,000,000
Events 162,000,000 200,000,000 14,580,000,000 18,000,000,000 64,800,000,000 80,000,000,000
GPS Updates (Loads x 600) 1,620,000,000 6,000,000,000 145,800,000,000 540,000,000,000 648,000,000,000 2,400,000,000,000
Invoice 2,700,000 243,000,000 1,080,000,000
IoT
Rates
Driver Pay, Settlements
Integration
Others
Total 2,052,800,000 6,860,000,000 184,200,200,000 587,585,000,000 818,646,200,000 2,610,335,000,000
Daily Transaction OLAP (2 yrs @ 200days/yr)OLTP (90 days)

5. Fragmentation of Transportation
Domain Expertise
Freight Volume
Regional Dominance
Advantageous Pricing
Historical Factors of a Fleet’s Success
 Effects
– Very little market-wide data available
– Every fleet has their own “secret sauce” they protect at all cost
– Brokers have access to the most data points

6. The Small and Mid-Size Carrier Disadvantage
 A daunting deficit of data available to the small carrier
 Primary way to overcome is to price low
– Even “low pricing” is defined by a best guess
3
30
100 Truck Carrier
2000 Truck Carrier
Avg. Company Age (years)
5000
50000
100 Truck Carrier
2000 Truck Carrier
Avg. Daily Loads Available
1,000,000
300,000,000
100 Truck Carrier
2000 Truck Carrier
Avg. Historical Lane Data Points Available

7. The Opportunity and Use Case

8. Small Fleets
Etc
The Commonality
TMW
Rate
Miles
Cost
Fuel
HOS
Breaks
Weight
Large Fleets
Etc
Rate
Miles
Cost
Fuel
HOS
Breaks
Weight
Medium Fleets
Etc
Rate
Miles
Cost
Fuel
HOS
Breaks
Weight
 No matter the fleet size,
thousands of them have
one thing in common when
it comes to lane data……..
 A TMW Database

9. Leveling the Playing Field
Through MRI, any Fleet of any size has access to:
 Average Rate
 Minimum Rate
 Maximum Rate
Which gives them the ability to:
 Negotiate better rate structures with shippers
 Identify those lanes in your network that can easily bear a rate increase
 Analyze new lanes and move into lucrative new markets
 Quote new business to win with competitive intelligence on current rates
 Reallocate internal resources to work more profitable lanes
 Accessorial Revenue Per Load
 Fuel Surcharge Per Mile
 Geographical Level
 Standard Deviation
 Number of Carriers
 Number of Loads

10. Technical Solution

11. SaaS Solution In An On Premise World
 Multi-Tenant Hortonworks
Hadoop Business intelligence
Solution Built in the Azure
Cloud
 TMW customers enabled with
Big Data and Data Science
without the cost of
infrastructure and personnel
 TMW Business Intelligence and
Data Science teams providing
solutions for customers and
industry

12. Components
Data Access Governance Operations Security
Hive 1.2.1/TEZ 0.7 Atlas 0.7.0 Ambari 2.4.2 Knox 0.9.0
Spark 2.0 Ranger 0.6.0 Oozie 4.2.0 Ranger 0.6.0
Hbase 1.1.2 Talend* Kerberos
Phoenix 4.7
Zeppelin 0.6
SSIS*
NiFi 1.1.0
*Not part of Hortonworks Data Platform

13. Building A Transportation Data Community
 The TME Data Community
enable Analytics and
Benchmarks for
transportation industry
 TMW is able to leverage
Data Science and Statistical
approach to Market
Intelligence for data
profiling and cleansing

14. Data Science For Transportation

15. MRI Visualizations
 Facilitates telling and sharing stories about data.

16. Fuel Data Analysis
 White paper for analyzing fuel prices and providing
evidence towards the value that ExpertFuel provides.

17. K-Means Clustering And Beyond

18. Model Building
 Rate per Mile,
 Load Availability,
 Load Velocity.

19. Model Saving/Loading & Predictions
 Enable Prediction As A Service by creating an on demand
model library.

20. Transportation Data Lake For the entire Industry

21. What Went Right

22. SaaS BI Successes
 Building the Data Lake and working in Hadoop in Azure
allowed the data lake to feed innovation in Predictive
Analytics, Master Data Management, and Benchmarking
 Use the experts Like Horton’s PS team(teach, teach, teach)
 Individual Customer and Industry Intelligence was secured
and centralized for collaborative Business Intelligence
 Speed to Market, TMW had it’s first customer on-boarded
in under 9 months, with more quickly added.

23. Lessons Learned

24. Lessons Learned
 Build a bigger data lake, the TMW
Data Community should have been
an early target to launch
 Be Agile in your R&D processes to
enable changes in process, big data
is changing fast with additions like
Apache Spark, Zeppelin and NiFi