Predictive Analytics in Supporting Logistics Operation Decision Making

Alhamdulillah. 


Please Enjoy, our Decision Support Systems article (co-authored with Dr. Andries Stam (Almende BV) and Prof. Eric van Heck (Rotterdam School of Management, Erasmus University).


The article's title is "The seaport service rate prediction system: Using drayage truck trajectory data to predict seaport service rates". It gives a concrete example on the use of predictive analytics using large volume GPS data (big data) to support logistics operational decision making (in the context of containers pick up and delivery operations at seaports. Click the following URL or feel free to contact me at meditya.wasesa (at) gmail.com for the full article: http://dx.doi.org/10.1016/j.dss.2016.11.008

Cite the article as follows:

M. Wasesa, et al., The seaport service rate prediction system: Using drayage truck trajectory data to predict seaport service rates, Decision Support Systems (2016), http://dx.doi.org/10.1016/j.dss.2016.11.008  

Business Process Re-Engineering (Photo Blog)

The Exact Definition of Logistics & Supply Chain Management

Many people tend to associate the logistics term and supply chain management term as synonymous.  But, are they really refer to the same concept?  In response, I try to cite some statements from a handful of academical papers to confirm this issue.  Enjoy!!!

- my kid's version of supply chain management -

"Logistics Management: the process of planning, implementing, and controlling the efficient, cost-effective flow and storage of raw materials, in-process inventory, finished goods, and related information flow from point-of-origin to point-of-consumption for the purpose of conforming to customer requirements" (Cooper et. al. 1997).

"Supply chain management is the integration of business processes from end user through original suppliers that provides products, services, and information that add value for costumers."  (Cooper et. al. 1997).

“SCM represents one of the most significant paradigm shifts of modern business management by recognizing that individual businesses no longer compete as solely autonomous entities, but rather as supply chains (consisting of individual businesses, working together).” (Lambert and Cooper 2000)

“The term SCM [is] used to explain the planning and control of materials and information flows as well as the logistics activities not only internally within a company but also externally between companies” (Chen and Paulraj, 2004).

“Supply chain management (SCM) encompasses the planning and management of all activities involved in sourcing and procurement, conversion, and all logistics management activities.  Importantly, it also includes coordination and collaboration with channel partners, which can be suppliers, intermediaries, third party service providers, and customers.  In essence, supply chain management integrates supply and demand management within and across companies”, according to the Council of SCM Professionals (Moonen 2008).

"A review of the SCM literature reveals that confusion exist in terms of what SCM actually is.  Nevertheless some commonalities do seem to exist:

•  It involves through several stages of increasing intra- and inter-organizational integration and coordination: and in its broadest sense and implementation, it spans the entire chain from initial source to ultimate consumer.
•   It potentially involves many independent organizations.  Thus managing intra- and inter-organizational relationships is of essential importance.
•  It includes the bidirectional flow products (material and services) and information, the associated managerial and operational activities.
• It seeks to fulfill the goals of providing high costumer value with and appropriate use of resources and to build competitive advantages." (Cooper et. al. 1997).

- end of 2013 - may we always be better persons in many years to come  -

 
References

Chen, I., & Paulraj, A. (2004). Towards a Theory of Supply Chain Management: the Constructs and Measurements. Journal of Operations Management, 22(2), 119–150. 

Cooper, M. C., Lambert, D. M., & Pagh, J. D. (1997). Supply Chain Management: More Than a New Name for Logistics. International Journal of Logistics Management, 8(1), 1–14.

Lambert, D. M., Cooper, M. C., & Pagh, J. D. (1998). Supply Chain Management: Implementation Issues and Research Opportunities. International Journal of Logistics Management, 9(2), 1–18.

Moonen, H. (2009). Multi-Agent Systems for Transportation Planning and Coordination. Erasmus University Rotterdam.

Object Oriented VS Agent-Based Software Engineering

Found a nice explanation that points out the difference between object-orientation and agent-based software while I was reading a thesis entitled "Multi-agent Systems for Transportation Planning and Coordination" written by Moonen (2009).  

He wrote, "From a software engineering design perspective, it is good to understand where agent-based approaches differ from traditional Object Orientation (OO) development methods. ...

Jennings (2001) listed the most compelling differences between agent-based and OO: 

•  Objects are generally passive in nature – they need to be send a message before they become active;
• Objects do encapsulate state and behaviour realisation, they do not encapsulate behaviour activation (action choice) – more specifically, an agent can have behaviours which are reactive, proactive, and/or social in nature;
• OO fails to provide an adequate set of concepts and mechanisms for modelling complex systems;
• OO approaches provide minimal support for specifying and managing organisational relationships;
• Agents have at least one thread of control but may have more, whereas Objects have solely one thread of control (Wooldridge, 1999)."  

little robot by annares - source openclipart.org

 

Just exactly what I have been looking for, maybe it is the case for you too ;)

References

Jennings, N. R. (2001). An Agent-based Approach for Building Complex Software Systems. Communications of the ACM, 44(4), 35–41.

Moonen, H. (2009). Multi-Agent Systems for Transportation Planning and Coordination. Erasmus University Rotterdam.

Wooldridge, M. J., & Jennings, N. R. (1995). Intelligent Agents: Theory and Practice. Knowledge Engineering Review, 10(2), 115–152.

Say It with Plots

Credits to R, GGPlot2, and Cookbook for R Blog.

Photo Blog: Neural Network Prediction Model

Documenting some old notes about the formalization of the neural nets prediction model.

I think the note about the "classic" Levenberg-Marquadt rule for dataset decomposition (into into the training, the validation, and the testing sets) also deserves some highlight.

Reviewing the results of the notes documentation I did with my scanner (Canon Pixma MP 540) (see the software release management blog) and my camera (Nikon D 5100) (see this blog),
I think I will go on using the camera instead of the scanner due to the picture quality, practicality, and blog aesthetic issues :)

Software Release Management - Scanned Notes

Conserving some knowledge from my old note book should not be a bad idea...

 

... for some notes, it's better than throwing them directly to the trash can.

Reviewing Machine Learning (Prediction Model) Diagnostics

Suppose, you've done all you need to do to build a prediction model.  Despite your hard work and high expectation, the prediction model you've designed performs poorly.

What to do next then?

After having a discussion with a good friend, he suggested me to review Andrew Ng's advices for debugging predictive analytic (machine learning) model.  He referred me to the video series below:



Having all the videos in the series (Advice for Applying Machine Learning - What to Do Next 1-7) reviewed , I learned about the machine learning diagnostic issue.  To Andries Ng's words, machine learning diagnostics is defined as: "a test that you can run to gain insight what is/isn't working with a learning algorithm and gain guidance as to how best to improve its (predictive) performance." 

Machine learning diagnostics is built on the identification of "bias" and "variance" symptoms of predictive models.  Based on the symptom's identification process, we can then define the prospective measures that will likely lead to better result.

So, what to do next then?
Based on the advice from lecture sets, the alternative measures are:

 ... and back to work, for sure :p 

Free Statistical Machine Learning Books and Lecture Slides

Statistical Machine Learning - Data Mining - Business Intelligence Books.

Inspired by recent David Smith's blog post about the possibility to downloading free Statistical Machine Learning e-book, I try to "re-echo" and "re-package" the exciting news.  Thanks to the generous researchers who make their valuable works freely available, we can download free digital copy of their books (either whole or as subsections) and the accompanying lecture slides.  Herewith, I list the URLs for downloading some data mining/ predictive analytic/ statistical machine learning resources

1.  "Elements of Statistical Learning (Full Book)" by  Trevor Hastie, Robert Tibshirani, and Jerome Friedman.

2.  "Information Theory, Inference, and Learning Algorithm (Full Book)" by David MacKay.

3.  "Forecasting: Principles and Practices (Full Book)" by Rob J Hyndman and George Athanasopoulos.

4.  "Introduction to Data Mining (3 Chapters of the Full Book and Complete Lecture Slides)" by Pang-Ning Tan, Michael Steinbach, and Vipin Kumar.

5.  "Machine Learning (Lecture Slides)" by Tom Mitchell.

6.  "Pattern Recognition and Machine Learning (one Sample Chapter and Some Lecture Slides)" by Christopher Bishop.

7.  "Data Mining for Business Intelligence (Instructor Materials)" by Galit Shmuelli, Nitin Patel, and Peter Bruce.

(the following URLs do not really fits in the data analytics topic, but they're useful information to share anyway:D).

8.  "Multiagent Systems - Algorithmic, Game-Theoric, and Logical Foundations (Full Book)" by Yoav Shoham and Kevin Leyton Brown.

9.  "Fundamentals of Multiagent Systems with NetLogo Examples" by Jose M Vidal.

10.  "Essential of Metaheuristics (Full Book)" by Sean Luke.


Hopefully the URLs will still work in the future,  Enjoy then.. :)

Knowing Extra Data Analytic Tools can be Handy

In the last period, I have been using R for preparing and analyzing big sized data.  It has been a reliable tool for many of my needs.  The existence of numerous blogs (e.g. r-bloggers) that share many useful tips makes trouble shooting can be done reasonably fast.

Memory Allocation Notification - R

Not without flaw, I just realized that R does have its limit.  At the moment, it can not allocate working memory more than 2.2 GB regardless how powerful your computer is (see the description picture above).  Although you have 8 GB RAM, R will send an exception when the utilized memory reaches the allowed limit.  To solve this issue (i.e. coping with "big data" analysis using R) Ryan Rosario's video can stand as a good reference.

EM Clustering - Weka

I haven't really experimented with what the video has suggested (I am not in the mood for learning new package :p).  In exchange, for my problem at the time (doing EM clustering for ~25,000 records data), I just used my old "fellow" Weka, and solve the problem within 30 minutes or so.

Because a tool's limitation can be explored at unexpected time, it is always handy to know more than one statistical learning tools (e.g. R, Weka, Rapidminer, etc).  Happy data crunching then :).  

Tesco Korea's Unique eCommerce Approach

Introduction to RFID, Nice 4,5 Minutes Video

Credits to Explania for this brilliant animation. Enjoy!!!

Nice Thesis -> Amazon and Wal-Mart's Supply Chain Practice

If you are looking for a good detailed reference on how big retail industries are executing their supply chain, this MIT master thesis doc (2005) may shed a light.  Colby Ronald Chiles and Marguarette Thi Dau have written quite an elaborate case study on how Amazon and Wal-Mart arrange their logistics.  You can download the document via the following link. 

openclipart.org / montacarichi

Have fun then :)

When AI and Mechanical Engineering Meet.

Fascinated by the AI and the multi-agent systems concepts in my present academical journey, I begin to see clearly that my previous academic background (bachelor in mechanical engineering and master in logistics engineering) is actually coherent with the existing skill set that I am currently exploring.

The KIVA robotic system, a marriage between the multi-agent system and mechanical engineering fields, has really opened my eyes.




Guess what, one of the KIVA co-founders, Raffaello D'Andrea, is a control systems (mechanical) expert.



Tried to browse another videos of him and found the following video from the cmurobotics.  CMU robotics is really becoming one of my most referred youtube channels at the moment.



Stay hungry, stay foolish (Steve Jobs), stay researching... (Meditya Wasesa) LOL :p

R & Gephi - a Starter Story

For the last couple of weeks, I have forced my self to use R-statistics for my data mining / machine learning/ predictive analyses.  The big fear was coming from the basic freaky "geeky" console that does not reflect user friendliness.  However, someone recommended me to use R-Studio for a better way to access R.  I've used it and have become a bit addicted to it.

The growing R communities that have developed numerous useful free packages thrill me a lot.  We can find a lot of free packages for any data analysis technique.  We just have to invest a bit of time to read each package's documentation, try-out some easy tutorials, and then apply the technique to our problem.  Moreover some visualization packages that will convert your analysis results into appealing charts are also available (e.g. the legendary "ggplor2", arulesViz, etc).  I show a sample visualization in the chart below, for more great stuffs and research inspirations we can visit the r-blogger and many other great blogs.

So why wait using this brilliant powerful free- analytic tool.  One other thing, in the near future I may need to visualize network charts beautifully.  The gephi free software may be a strong option.  I've noticed that many (network) researchers recommend it, I may give it a try (if the time permits.. help..@_@) .. The video teaser is quite appealing...


Introducing Gephi at JavaOne from gephi on Vimeo.

Converting UNIX <-> Human Readable Timestamps in R and Excel.

For the last two days, I have been trapped in a tricky problem (for a beginner like me :p); converting UNIX timestamp to human readable timestamp.

To solve the problem, firstly I went to this site  http://www.epochconverter.com/ .  I entered a sample UNIX timestamp record which is “1333456742556”.  The site successfully converted the data into “GMT: Tue, 03 Apr 2012 12:39:02 GMT”.  Hmm quite reasonable..

Then to convert the whole dataset I tried the code recommended by the site.

for (R) <- as.POSIXct(theUNIXtimestamp, origin="1970-01-01")

for (Excel) <- (theUNIXtimestamp / 86400) + 25569

*note you have to adjust the time with your local timezone, (see the URLs at the end of this post).

When I applied the code, one funny thing happened.  The code convert “1333456742556” into "4225-07-25 03:42:36 CEST".  How could it be?  What’s wrong?

After sometime, I find the answer.  The key is to use the first 10 digits of the UNIX code and remove the remaining digits.  Note that the original UNIX timestamp, “1333456742556” consists of 13 digits.  Thus, I extracted the first 10 digits (“1333456742”), … re-used the code, and … abracadabra.. “"2012-04-03 13:39:02 CEST".. the correct timestamp appear..  HURRAY..  (beginner’s excitement hehehe..)

Credits for these URLs ( those lovely people who spend a bit of their time to share their knowledge.. :) )...

(for General Issue) 

http://www.epochconverter.com/

(for Excel) 

 http://spreadsheetpage.com/index.php/tip/converting_unix_timestamps/

 http://www.mrexcel.com/forum/showthread.php?t=130691

(for R)

http://stat.ethz.ch/R-manual/R-devel/library/base/html/as.POSIXlt.html

The Cheapest and the Most Expensive Export Destinations

To get an impression about the cost magnitude in exporting a container to different countries, we can browse the survey site provided by the world bank (this URL). Based on the 2010 data, the 5 cheapest export destination countries are Singapore (450 USD), Malaysia (456 USD), China (500 USD), United Arab Emirates (521 USD), and Finland (540 USD) and the 5 (6 actually) most expensive export destinations are Chad (5902 USD), Central African Republic (5491 USD), Republic of Congo (3818 USD), Iraq (3550 USD), and Afghanistan & Nigeria (3545 USD). In addition, exporting a (20 TEU) container to my homeland (Indonesia) will cost you (approximately) 644 USD, exporting a container to US will cost you 1050 USD, to Holland 895 USD.  

taken from openclipart.org

Mmm.. Oyeah.. to get some practical (in complementary with the academic one :p) knowledge about the exporting formalities and commons, the website from the US  http://export.gov/index.asp can be considered as a good starter. Okie.. have fun then.. ;)

International Trade Incoterms

When one tries to understand the global logistics processes, she has also to be familiar with the ICC (International Chamber of Commerce)'s incoterms condition. For a brief understanding, we can find sufficient information from the wikipedia (the English page is more up to date, but the Dutch page has a nice descriptive picture). "Incoterms (International Commercial terms), a series of three-letter trade terms related to common sales practices, the Incoterms rules are intended primarily to clearly communicate the tasks, costs and risks associated with the transportation and delivery of goods (wikipedia)". 

 Incoterms (wikipedia.nl) 

Have a nice holiday then.. :)

Dynamic Vehicle Routing - PhD Theses Couple

It is interesting to see, how some PhD candidates have managed to develop such a good team work in doing research and produced articles that are mutually beneficial for each team member's thesis. I find this unique case while I am drowning in the dynamic vehicle routing problems literature. Basically there are two approaches for tackling the problem: the OR (operations-research) approach and the decentralized (agent based) approach. While Tamas Mahr have concentrated on developing the agent based solution, Jordan Srour have invested some time in tackling the problem from a centralized perspective. They collaborated, produced a great paper (Transportation Research paper), and have a nice chapter in their theses. Nice..

Comprehensive Introduction to Cloud Computing & Multi-tenant Architecture

Hi, I have been exploring the internet for some useful material about cloud computing, SaaS, and multi-tenant architecture. For a novice like me, most of the articles out there are either too "geeky/ technical" or too "pop". In the end, I finally find a nice source of information from Microsoft's MSDN Library. Actually, I was expecting something better from the Google's side, but the URL I found is just too technical for me.

If you go to this URL, you can find some articles which are not so low level, yet not too high level either. Take an example, the article "Architecture Strategies for Catching the Long Tail" explains the business rationale of developing multi-tenant SaaS application (i.e. the relationship between the deployment instances and the economics of scale). Another brilliant article entitled "Multi-Tenant Data Architecture" explains nicely the issues that have to be considered in developing multi-tenant application (i.e. user management authority, data model customization for different tenants, etc). Surely there are a lot of things to explore in the mentioned MSDN library. The articles are written in 2006, and I just explored them now, what a laggard :). Anyway, in this case (transferring the SaaS - multitenancy issue and presenting them in a digestible manner to non "geek" people), I find that Microsoft is better than Google. Well, have a nice weekend everyone.