Murphy Choy

Archive for October, 2011|Monthly archive page

Business Analytic Opportunity Identification Process: Ontological Approach

In Uncategorized on October 23, 2011 at 6:18 am

A few days ago, we were having a meeting and discussing about the aspects of coming out with a procedure to identify processes or functional area groups where analytics can be applied vigorously and regularly in order to optimize the business. Now this is a very good idea in theory, but it is very difficult practice in reality. One of the key issue that one will face in trying to come out with such a process is that one can never truly understand all the underlying processes that governs an organization. This is particularly the case for companies that are very big all or multi-national in nature.

However this does not mean that we cannot identify processes in order to improve them. The simplest thing that one can actually do is to actually map out the entire process or the entire operational model or aspect of the organization. However, this is not without its difficulties. Many organizations because of the long established history of operation, they have actually forgotten about the origins of the many existing processes and could not give satisfactory explanation as to why this processes were placed or implemented. This resulted in many confusing meetings in which many senior managements will have difficulty communicating with the middle managers.

In order to identify processes, there are a few approaches to doing so and the approach given above is just one of the many processes possible used to identify potential areas. Now the above process is actually an empirical process which relies on the ability to map out the processes of the company. However in this case, I propose an ontological approach towards the identification of potential areas where business analytics can be applied easily. In in my proposal, suppose that any processes or any functional area in which it can be measured quantitatively – they can be optimized further through the use of analytic. Why do I use quantifiable measures? The reason behind is only quantifiable measures in our organization can be optimized by understanding the constituent components, modeling them and in turn improving the overall results. This is far more difficult to achieve using qualitative information. While this is something that I have not experimented with existing data, because this is an ontological approach this is after all a theoretical move.

A dangerous aspect of analytic: dark side of the force

In Uncategorized on October 14, 2011 at 5:10 pm

I was having some discussion with fellow colleagues about the applications of analytic in public sector. We did appreciate some of the great improvements that have occurred as a result of the various analytic initiatives. However, there were some very interesting observations about how analytic can be abused when money is concerned.

We were having a discussion about public amenities and some one in the crowd mentioned that some public services have stopped offering services to people who cannot pay. Now it is useful to understand that public services are not designed to be profitable business and thus non-paying customers are serious problem for them. To prevent this, they have adopted some kind of analytics to prevent this from occurring through rejection of people who are deemed to be unable to pay.


Public services existed to serve the greater public and ensure that the public has access to these amenities. It is almost unthinkable that they actually abuse analytic to deny public access to amenities. Analytic as with all sciences was developed in hope of helping business make better decisions and achieve higher performance but never at the expense of public interest.The other issue is that analytic works on the basis of probability. What if you classified someone wrongly and thus denied that individual of rightful access to amenities?

I sincerely hope that public sectors will look to it as their duties to serve the public and not abuse analytics to deny people of their amenities.

Three different ways to do forecasting in SAS

In Uncategorized on October 12, 2011 at 3:54 pm

Well, we are having a discussion about the tools available for forecasting purposes. The reason why we are discussing this was because we were evaluating the various possible ways to obtain forecasting for a possible project with a company. as the company currently do not have the essential tools to meet their forecasting purposes, they need our help to recommend a suitable tool.


Since I am one of the great fan of SAS, I have recommended a few possible choices from their menu of choice. The first tool to recommend is the SAS/ETS product. This is the choice for people who are comfortable with base programming. At the same time, it offers a great variety of possible models for analyst who might be facing forecasting problems or Time series issues.


Another possible choice is to use the forecasting studio which is configured to use high-performance forecasting technologies that are embedded within the SAS technologies. The forecasting studio is a powerful tool and has extended forecasting capabilities above that of SAS/ETS. While it is a powerful one, it is also more costly than the base version.


Finally, the last tool for forecasting purposes is JMP. While it is not technically a part of SAS, it is one of products of SAS. It has very graphical approach to time series analysis and provides a variety of tools to achieve accurate forecasting.

Migrating data in SAS: 32-bit to 64-bit

In Uncategorized on October 11, 2011 at 3:35 pm

Today, I was having difficulties in reading some of the datasets. At first, I was thinking that the error might be due to incompatibility between type encoding. This has happened before in some of the other datasets that I was involved in. However, it turned out to be an issue because of the native system that is in different bit.


Most fortunately, SAS provides PROC migrate which allows us to transport data sets between systems. While this is not a panacea to all our data set problems, it does alleviate some of the issues that we face. At the very least, the conversion process is a lot more simpler and easier to manage.

Singapore data protection act: it’s impact on business

In Uncategorized on October 10, 2011 at 3:56 pm

Recently, the Singapore government has decided to review its existing data protection schemes in which there were attempts to change the existing model from opt in to opt out structure. This sudden change will have a dramatic effect on the local business. This is because the act is going to impact the abilities of the local businesses in sharing their data among businesses. While it is harmful to the local business, it will ultimately benefit the entire economy.


One reason why the data protection act is so important is because it allows the consumers to protect their private information from users or businesses whom they have not authorised to access this information. While many will argue that information is easily available from the Internet or lucky draw surveys, one must know that when the new regulation comes into force it will prevent collectors of such information from selling this information to third parties. This act will ultimately prevent the proliferation of mystery calling where scammers will attempt to extort money from innocent parties. I believe this is a direct response to the widespread and increasing incidents of calling extortion cases originating from overseas sources. while there are alternatives to extract information, these channels will be by far few and limited in depth.


However, data protection act will benefit the economy as Singapore is growing as the regional data centre as well as analytic hub. Such strong regulations as well as preventive measures will indirectly cut any abuses of data and promote Singapore as a positive data destination for countries in Europe and the United States of America. While most of the analytical work are happening in India or China, countries such as America will find it’s sensitive to send information to anywhere in China. Singapore being a neutral place will be the ideal destination as a data centre hub for many of these countries.


Hopefully this act will ultimately benefit the Singapore economy.

Worms in your meal and operational risk

In Uncategorized on October 7, 2011 at 7:00 am

Many of us would have experienced eating takeaway. Because the food has to be prepared in a manner that makes it easy to pack and cook,very often shortcuts are being taken to achieve this. One of the many shortcuts is to reduce the amount of time taken to wash vegetables. This is often the cause of worms and unwanted pests and bacteria in the food.

Very often most of the food producers or vendors concerned this to be part of operational risk. Unfortunately, food hygiene remains low on the priority list. Because of the huge amount of customers, some of these vendors could afford to have shady reputation. Hopefully with the rise of social media, we would be able to identify the black sheep.

Some thoughts about operational risk

In Uncategorized on October 5, 2011 at 7:00 am

A few days ago, there was a seminar about operational risk which covers the basics of operational risk management as well as some of the challenges faced by the practitioners. During the entire session, there was extensive coverage of the self risk control assessment. It was an interesting discussion and this is in turn used by some of the students in the course work.

There was a student who discussed about the use of e-mail surveillance in order to analyse the content of e-mails so as to create an early warning system for the management. This is an interesting idea but nonetheless a dangerous one. In fact this is one of the key issue debated by the proponents of liberty against those of control. And I can understand the dangers and possible misuses of the system. In fact this is one of the reason why an ex-CEO was removed from his position.

While there have been extensive research into masking of data, it is still possible to uncover the identities of individuals and likely to lead to privacy issues. And this is one of the operational risk issue that one may face when they are trying to build a system to highlight issues in the management. Careful use of data and elaborate systems will be needed in order to paint the proper picture of operational issues in the company without infringing the privacy of the employees.

Cable gate data and opportunities

In Uncategorized on October 4, 2011 at 2:39 pm

I was having a short discussion with my fellow researchers who specialise in text analytic about the potential of the cable gate data. However, they have much reservation about the usability of the cable gate data in general. Much of the worry rests on the sensitivity of the data as well as its reliability.


In my opinion, there are two major areas that we can research on. The first area that we can look into is the research on document similarity analysis. This analysis will be useful to identify documents which are written by the same author. This is useful for several reasons such as identification of the number of unique sources as well as the reliability of the data. Of course, this is no easy feat but there is huge potential that we cannot ignore.


The second area of analysis would be to identify the lexical structures within the document which is so often ignore by so many researchers. This allows us to build factual lexical identification algorithms that might prove to be useful in many other situations.


Due to the sensitive nature of the data, it will be very unwise to look into the contents which might in some way or the other compromise the security of another nation’s people and interests. However, structures do not point to any information that are directly related to something which we can identify easily from existing information.

Speech recognition blogging

In Uncategorized on October 1, 2011 at 6:31 am

It is a gentle reminder. All my existing blogs are updated by the use of Dragon the speech recognition software. You’ll be surprised how simple it is to use this speech recognition software. While it is not the best, it is most likely sufficient for most work. Thank you.

Satirical comments and text analytics

In Uncategorized on October 1, 2011 at 4:33 am

Recently there has been comments and criticisms of text analytics being incapable of managing, analysing and interpreting blogging sites, Facebook pages, web forums and tweeting online. This is especially the case for webpages which contain linguistic style that are too flamboyant and satirical in nature. All these comments are done in good will but they are misguided.

Much of the problem analysing these problems relates to the difficulty in developing an appropriate corpus. Very often we are limited by the context of the corpus material being used. However, these materials tend to be generic in nature and thus cannot be expected to perform to its maximum potential. The other issue is that they are actually limited in terms of scope by the material that are commonly available.

The fastest and simplest solution to this uncommon problem is the active use of customisation in the development of corpus as well as a good dictionary of  idiom. At the same time, you could also train your system on the information provided as part of the training base.