Big data in freight transport

(1)

Big data in freight

transport

!

Per Olof Arnäs

Chalmers

@Dr_PO

[email protected]

!

Slides on slideshare.net/poar

(2)

We are in the middle of a gigantic

exponential

development curve

(3)

”Fast Up-and-Coming

Movers Toward the Peak

Are Fueled by Digital

Business and Payments”

”…the market has settled

into a

reasonable set of

approaches

, and the new

technologies and practices

are

additive to existing

solutions

”

(regarding the decline of Big data on the curve)

Gartner, August 2014

(4)

Gartners Hype Cycle for Emerging Technologies

(5)

Gartners Hype Cycle for Emerging Technologies

(6)

So…

What is

Big data?

(7)

2011 2013 2015

”Big data is an

all-encompassing term for

any collection of data

sets

so

large

and

complex

that

it becomes

difficult to

process

using on-hand

data management tools or

traditional data

processing applications.”

- Wikipedia

(8)

Google flights

(9)

Jawbone measures sleep

interruption during earthquake

(10)

smile! by Judy van der Velden (CC-BY,NC,SA)

Speculative

shipping

http://www.scdigest.com/ontarget/

14-01-21-1.php?cid=7767

(11)

http://www.scdigest.com/ontarget/

14-01-21-1.php?cid=7767

Speculative

shipping

Package item(s) as a package for

eventual shipment to a delivery address Associate unique ID with package

Select destination geographic area for package

Ship package to selected distribution

geographic area without completely

specifying delivery address

Orders

satisfied by item(s) received?

Package redirected?

Determine package location

Convey delivery address, package ID to delivery location

Assign delivery address to package Deliver package to delivery address

Convey indication of new destination geographic area and package ID to

current location

Yes

Yes No

No

(12)

Not statistics

Exhausted by Adrian Sampson on Flickr (CC-BY)

(13)

Not

Business

Intelligence

Basingstoke Office Staff Desk "No computer" by John Sheldon on Flickr (CC-BY,NC,SA)

(14)

(15)

Business

processes

Infra-structure

Paper based Phone   Pa_per s Road signs Analogue tools RDS Monitor fuel cosn umption Digitization version 0 0.5 1.0 1.5 2.0 E-mail Fax TMS-sy stems _Ex cel Route planning GPS f or na vigation Electr onically g ener ated freight documents Barcodes RFID-ta gs Simple or der handling Advanced or der handling Open interf ace W eb based UI Pla tform based sy stems Har dw ar e-oriented Da ta collection sy stems (pr oprietary) Comm unica tion with vehic les E-in voice We_{b based} book ing Route optimisation T_{he social w} eb Open connectivity Inte gra_ted pr ognosis Da ta collection systems (open) Tolling systems

Webser_traffvices with ic data

Dynamic _routing systems P erf ormance Based access Perf ormance Based access Mashups   Multiple da ta sour ces Probe da ta Individual _routing informa tion Pla tooning Platooning Exce_ptions handling Smar t goods Manual Computers Software Functions Distrib uted decision _mak ing Goods as bi-directional hyper link Paper based

CC-BY Per Olof Arnäs, Chalmers

Goods

_Vehicle

Barcodes RFID Sensors ERP systems TMS systems E-invoices Cloudbased services Order handling Driver support Vehicle economics RDS-TMC Road taxes Active traffic support Pr edictiv e maintenance 2014-08-26

(16)

Strategic Tactical Operational Predictive

Time horizons

We are approaching

this boundary

…and we are

starting to

move past it!

Real-time!

(17)

En la cima! by Alejandro Juárez on Flickr (CC-BY)

(18)

3 data types

Mountaintop #1

Collection of data in real-time

(19)

Mountaintop #1

Collection of data in real-time

5 data domains

Vehicle

Driver

Cargo

Company

Infrastructure/

facility

at

lea

(20)

Length  Weight  Width  Height Capacity  + other PBS-criteria Emissions  Fuel consumption  Route Position  Speed  Direction Weight  Origin  Destination Accepted ETA Temperature 

+ other state variables

Temperature + other state variables Education/training Speed (ISA)  Rest/break schedule  Traffic behaviour Belt usage 

Alco lock history

Schedule status (time to next break etc.)

Contracts/

agreements Previous interactions Backoffice support

Fixed

Historical

Snapshot

Vehicle

Cargo

Driver

Company

Infrastructure

/facility

+ fixed data layersMap Traffic history

Current traffic Queue

Availability

(21)

Mountaintop #2

Processing of data in real-time

(22)

Mountaintop #2

Processing of data in real-time

(23)

Mountaintop #3

Exploiting data in real-time

Connected. 362/365 by AndYaDontStop on Flickr (CC-BY)

Lisa for I/O Keynote by Max Braun on Flickr (CC-BY)

Fulham-Manchester United

24-02-2007 by vuhlser on Flickr (CC-BY)

(24)

Mountaintop #3

Exploiting data in real-time

(25)

CASES

 

(26)

CASES

 

(27)

Human resources

Reduction in driver

turnover, driver

assignment, using

sentiment data

analysis

Real-time capacity

availability

Inventory

management

Examples of applications in freight

(Waller and Fawcett, 2013)

Transportation

management

Optimal routing, taking

into account weather,

traffic congestion, and

driver characteristics

Time of delivery,

factoring in weather,

driver characteristics,

time of day and date

Forecasting

Waller, M. A. and Fawcett, S. E. (2013), Data Science, Predictive Analytics, and Big Data: A Revolution That Will Transform Supply Chain Design and Management. JOURNAL OF BUSINESS LOGISTICS, 34: 77–84

(28)

Manage complex systems

(29)

(30)

(31)

http://blog.digital.telefonica.com/?press-release=telefonica-dynamic-insights-launches-smart-steps-in-the-uk

(32)

7 Big Data Best Practice Across Industries

Usage of data in order to:

Increase Level of Transparency

Optimize Resource Consumption

Improve Process Quality and Performance

Increase customers loyalty and retention Performing precise

customer segmentation and targeting

Optimize customer

interaction and service

Expanding revenue streams from existing products

Creating new revenue streams from entirely new (data) products

Exploit data for: Capitalize on data by:

New Business Models Customer Experience Operational Efficiency

Use data to:

• Increase level of transparency

• Optimize resource consumption

• Improve process quality and performance

Exploit data to:

• Increase customer loyalty and retention

• Perform precise customer segmentation and targeting

• Optimize customer interaction and service

Capitalize on data by:

• Expanding revenue streams from existing products

• Creating new revenue

streams from entirely new (data) products

New Business Models

Customer Experience

Operational Efficiency

Figure 4: Value dimensions for Big Data use cases; Source: DPDHL / Detecon

2.1 Operational Efficiency

For metropolitan police departments, the task of tracking down criminals to preserve public safety can sometimes be tedious. With many siloed information repositories, casework often involves making manual connection of

many data points. This takes times and dramatically slows case resolution. Moreover, road policing resources are

deployed reactively, making it very difficult to catch criminals in the act. In most cases, it is not possible to resolve these challenges by increasing police staffing, as government budgets are limited.

One authority that is leveraging its various data sources is the New York Police Department (NYPD). By capturing and connecting pieces of crime-related information, it

hopes to stay one step ahead of the perpetrators of crime.6

Long before the term Big Data was coined, the NYPD made an effort to break up the compartmentalization of its data ingests (e.g., data from 911 calls, investigation

reports, and more). With a single view of all the informa-

tion related to one particular crime, officers achieve a more coherent, real-time picture of their cases. This shift has significantly sped up retrospective analysis

and allows the NYPD to take action earlier in tracking down individual criminals.

The steadily decreasing rates of violent crime in New

York7 have been attributed not only to this more effective

streamlining of the many data items required to perform casework but also to a fundamental change in policing practice.8 By introducing statistical analysis and

georaphical mapping of crime spots, the NYPD has been able to create a “bigger picture” to guide resource deployment and patrol practice.

Now the department can recognize crime patterns using computational analysis, and this delivers insights enabling each commanding officer to proactively identify hot spots of criminal activity.

6 “NYPD changes the crime control equation by the way it uses information”, IBM; cf. https://www-01.ibm.com/software/success/cssdb.nsf/CS/JSTS-6PFJAZ 7 “Index Crimes By Region”, New York State Division of Criminal Justice Services, May 2013, cf. http://www.criminaljustice.ny.gov/crimnet/ojsa/stats.htm 8 “Compstat and Organizational Change in the Lowell Police Department”, Willis et. al., Police Foundation, 2004; cf. http://www.policefoundation.org/

content/compstat-and-organizational-change-lowell-police-department

2.1.1 Utilizing data to predict crime hotspots

(33)

Domain

knowledge

critical!

See for instance: Waller, M. A. and Fawcett, S. E. (2013), Data Science, Predictive Analytics, and Big Data: A Revolution That Will Transform Supply Chain Design and Management. JOURNAL OF BUSINESS LOGISTICS, 34: 77–84

Data scientists -

the new superstars

"Data Science Venn Diagram" by Drew Conway - Own work. Licensed under Creative Commons Attribution-Share Alike 3.0 via Wikimedia Commons - http://commons.wikimedia.org/wiki/

(34)

Challenges

The Challenger by Martín Vinacur on Flickr (CC-BY)

Cross-disciplinary

Cross-industries

Cross-borders

(35)