Computer Aided Decision Systems – Industrial Practices Using Big Analytics

Professor Deepu Philip

Department of Industrial and Management Engineering
Indian Institute of Technology Kanpur
Lecture 01
Introduction to Decision Support Systems (DSS)
This is the first chapter of the decision support system course for practitioners and researchers
alike. As you might have read in the introduction slide, this course is meant for both
management practitioners, as well as students who are pursuing their career in management, in
specifically to the academic side.
So, we will look at both the applied side of it and a little bit of the academic side of it. So, it
will be a mixed flavour course. So, today is the beginning lecture and we will try to get into
this decision support system and why this decision support system is an important tool for this
management decision making.
(Refer Slide Time: 1:00)
So, if you see the slides, the title is Introduction to Decision Support Systems and my name is
Dr. Deepu Philip. I am from IIT Kanpur. So, in today's agenda, mostly we are going to talk
briefly about these concepts. We will enhance these concepts in the later presentations. The
first concept we talk about is the DSS concept. What is DSS and what does it entails to.
Then, we talk about slight history of DSS, how it actually progressed, how it actually spends
over time. So, this is the time historic progression. Then, acceptance of DSS: industry and
academia. We will be talking about the acceptances of the DSS in both areas.
Then, we talk about the uses of DSS. So, what, where and how the DSS is used, that are the
major aspects, we will be discussing on this. And then the last will be the introduction to. So,
we introduce major components, what are the major components of DSS is what we will be
discussing and then once this, we will move to the next one.
(Refer Slide Time: 2:26)
So, without delaying further, let us talk about the basics of DSS. As I said earlier, why is DSS
important and how does it come around? So, the DSS concept, there are so many people. So,
one of the questions is who started this concept. There are many debates on this, but widely
accepted name is Michael Scott Morton in 1970’s. He was the one who actually proposed this
concept or who created this concept, and the initial form it was known as management decision
system.
It was not called as decision support system, it was more called as management decision
system. The aim was to facilitate top management to take complicated business decisions. So,
this was the reason; the first version of it actually focused on the top management to make
complicated business decisions. So, hence the name was management decision system.
Then later it became, what we call as the decision support system. Then later as the computers
improved, computers started evolving faster. In 70’s, we were talking about mainframes. So,
later down the road in 80’s and 90’s we moved to what we call as personal computers or
powerful desktops.
So, as the computer system started evolving faster and more and more computation power was
available, this management decision system was characterized or told to us as an interactive
computer based system, a computer-based system that you can interact with. So, why do you
want to make it as an interactive computer system? This is the fundamental question of this.
The main reason why it has to be interactive is to help the decision makers. And how do or
how does the help happens? By utilizing data and models. And to achieve what? To solve
unstructured business problems. So, this word is very important, this unstructured business
problem is very important aspect in this one. Before we get into this, let us talk about the two-
type of problems.
As I said earlier, we can broadly classify the decision problems into two. One is called as a
structured decision problems and the another is the unstructured decision problem. So, what is
a structured decision problem? The structured decision problem is very well defined and second
thing is, it has complete data. Total data is available with you in this for this particular case and
the most important thing is, we know how to measure the efficiency of the decision. And most
of the time, we call it as also textbook problems. This structured decision problems are usually
available in textbook and very rarely available in real life.
In the unstructured decision problems or unstructured decisions, you need to apply rationality.
We will talk about what is rationality and rational decisions later down the road. You also need
to have some previous experience or a prior experience and there are other aspects to this. And
the third most important part is, there is no single decision.
So, we need to do alternate analysis. We need to look into different alternatives and we need
to find out which one of these alternatives is really effective and efficient in this regard. So,
most of the time, the unstructured business problems utilizes the experience. The other name
for experience is apriori knowledge. So, the unstructured business problems utilizes the
experience or the apriori knowledge of the decision maker.
Once the alternates are analysed by the decision maker, then you basically choose. You have a
set of options and from which you select the best alternative, whatever be the best one that you
think you can, that is the best alternative is the one that you end up choosing as part of this.
So, the idea of this one is, when you have to look in different alternatives, it will also give rise
to different views of the decision support system based on the background. So, the decision
makers background should also be factored; which means if the person is technically sound or
non-technically sound, can the person be comfortable with a computing system etc, that is all
part of the decision making at this point.
(Refer Slide Time: 12:15)

So, we now move into the definition of the DSS. How do we define DSS as part of this? So,
there are many definitions available. So, we are talking about practical or applied definition or
in other words about working definition.
1) It is a computer-based system.
2) It is capable of bringing together information from various sources. You have information,
you can call it as data also; when you process data you get information. So, either bringing
together data or information from various sources, either one.
3) Facilitate organizations or entities or individuals to analyse the information. It is not just
analysed. You can think about it as collect, classify, segregate and then group, etc. Analysis is
not just taking a numerical value out of it. You basically, you are able to do different actions
using this information.
4) Allow for the evaluation of assumptions and alternatives. So, what is assumption in this
regard? It is hypothesis or belief.
So, what is an example of hypothesis? Let us take a small example of a automotive tyre
manufacturer. The automotive tyre manufacturer believes that his tyre is superior to others. So,
the belief is superior tyre. Now, the superiority can be done in many ways. It can be on mileage,
so this tyre will run for like, let us say, one lakh kilometres or something, or it can be on safety.
It will never get punctured or something, or it can be on what you call efficiency. Minimize
tyre noise and other kind of things. So, there are so many ways you can establish the superiority
of the tyre. So, in different ways or different measures, these are all measurements or these are
all ways to quantify. So, system should allow you in multiple ways to quantify and evaluate all
these quantifications. That is one of the important aspects of this.
So, if any system that allows for all these four things, you can generally call it as, computer-
based system. It is capable of bringing to the information or data from various sources, and it
facilitate the analysis of information; not just calculating a numerical value but you can classify,
segregate etc. And it also allows for evaluating the assumptions and alternatives.
So, what does the DSS helps you to do? The DSS help us to improve. It helps you to improve
what? To improve the quality and responsiveness of decision making. What is quality? Quality
is conformance to standards.
In decision making, it is a conformance; but it is conforming to expectations. You make a
decision expecting something, and if whatever you expected happened as part of the decision
then you can call it as a quality, the decision as good quality. And responsiveness is the
quickness. How quick? The time that you take to make the decision.
So, a lot of the time people says there is a cost of ‘not making a decision’, which is usually
very, very high in corporate world. Not making a decision usually happens only in bureaucratic
world, not necessarily in corporate world.
So, when a system allows you to improve the quality and the responsiveness of the decision
making, what does you achieve? It allows you to improve the management of a corporation or
organization or an entity or a business and private limited company, whatever you want to call
it.
It will allow you to improve or better manage the organization. So, ideally speaking, the DSS
has evolved from EDP or TPS. TPS stands for transaction processing system. From there, it
went to something called as MIS. MIS is Management Information Systems. So, from TPS we
moved to MIS, from there it evolved to DSS.
One of the major reasons why this DSS evolution happened is, what happens about the data.
The data has become something like big data, divergent data, and so many data things come
into picture.
For us, in the decision support system course, we talk about data coming in two forms.
1) It is too much.
2) It is too little.
Either you have abundance of too much of data; otherwise you have too little of data. And in
either case, you give the too much of data, you are overwhelmed by the data, you have no clue
what to do with the data. And sometimes the abundance of the data just clogs up your decision
making and this is very much. This too much is usually seen in sensory, it results in sensory
overload.
And in too little, when you have very little of data, so, this is when you see people, pilots or
somebody who are flying an aircraft and when all these alarms are beeping. You can see that
they go into sensory overload and then they forget the fact that they have to just fly an aircraft,
at the some point of time. Then, when you are too little, what happens is, you make
assumptions; that you cannot verify, or irrelevant assumptions, and then make sub optimal
decisions as part of it.
So, the data whether it comes too much or too little, the decision support system is supposed
to handle both the scenarios or both the cases with equal capabilities in this regard.
(Refer Slide Time: 23:25)

Now, we move towards the next one; which is the first part of the decision support system, we
call it as electronic data processing, EDP or other name for it is, transaction processing system,
what we call as TPS. So, this was the initial form; the most primitive incarnation of DSS. You
can think about it as the father of DSS, or the beginning (starting) point of DSS. This is from
where the DSS evolution started.
It was intended to be applied to lower operational levels of an organization. So, let us take an
example of this. A bank, and one of the very lowest levels of operation is a bank teller or you
can call it as a cashier. So, you go to the bank, you take a cheque with you, or you take some
money with you and let us say you want to do a money deposit. So, you have something like
20,000 Rupees in cash with you and you want to deposit it. So, you go, you give the 20,000
rupees to the bank person. What do they do? They first count the money, and then they open
up the computer, they press a functional key, account number comes in, they type your account
number, put this much money is deposited, enter that and then the data goes into the system
and that is it. So, your transaction of depositing 20,000 rupees is facilitated by the system. The
main aim is to automate paperwork. Instead of you filling paper, they filling paper, signing,
sealing, giving it, or all those kind of things, this can minimize paper and human error.
In old days, in a bank, you had journal and ledger. So, what they used to do is, any transaction
on a day, they record it in two registers, then used to put in a ledger, then it would transfer to a
journal etc. By the time the bank operations are over by 2 pm or 3pm, until 6 PM you have to
reconcile all the paperwork.
So, now with the help of a computer, you do not need to do anything. All these transactions are
all digitally stored, and with a press of a button you should be able to find out what has
happened. So, all these kinds of things, what we call as the lower-level operational needs, this
do not require too much of intelligence. These are mostly the routine jobs. So, for facilitating
these kinds of routine jobs and minimizing paper and human error, we came up with the
electronic data processing and another name for EDP is also known as Canned Transactions.
So, this canned transaction means, they are pre-written computer programs; written, tested and
validated computer programs or application to process a transaction. Our indentation is one
transaction should be completely processed in that regard.
So, most of the bank database and other kind of things are classic example of electronic data
processing system. So, what are the main characteristics, how do you distinguish?
1) The operational focus of or on data. You are focusing more on data.
2) Then storage of this data. So, make sure that the data is stored properly.
3) Then processing of the data. So that, if there is a numerical value, addition, subtraction,
everything else need to be done and ensure that the complete data flow. So, they heavily
depend. So, that is why it is scanned, because in one press of a button, this application of the
program is available to you, and you can enter the necessary details and the transaction is
available for you. So that, you can immediately process the transaction. So, the transaction is
processed efficiently, that is the main part. So, most of the time it is functional key oriented. It
is scheduled and optimized computer runs. So, the CPU, input device, storage, output device;
this combination is scheduled and optimized in such a way that the complete transaction is
processed properly.
4) It integrate files for related jobs to make sure that all the files that are related to the job are
integrated at the same time. So, all those things are integrated.
5) It provides summary report for the management. So, you should be able to give the
management, what is the summary of the transaction. If we are talking about a bank, the bank
manager might need to know how much money was deposited, how much money was
withdrawn, how many 500-rupee notes, how many 1000-rupee notes, or we do not have any
1000-rupee notes, how many 2,000-rupee notes, or whatever it is. So, that kind of information
for the next upper-level management can also be provided as part of this electronic data
processing.
(Refer Slide Time: 34:32)

So, you started collecting more data on the transactions, and when you have more data on the
transaction obviously then, you want to find out more information out of the data. And from
that data you want to synthesize more information, which gives rise to the next revision of the
decision support system, what we call as the management information system.
Here, in the MIS, the focus is on information. So, are you just focusing on information? You
are focusing on information with an emphasis. You are emphasizing, with an emphasis on
integration and planning of the information systems function. So, in this case, what actually
you are looking at is, you may have an MIS, and MIS will be dealing with TPS1, TPS2, TPS3,
etc. So, you may have multiple transaction processing systems, which will all be providing
information to one MIS. So, may collect data from multiple TPS, and derive relevant, or
valuable, or necessary information, for the middle level management.
Most of the time, the MIS was originally intended. This middle level is a very loosely used
term in this regard. It is an aggregation of everything. The MIS is one level up, where multiple
transaction processing system data is collected in one go and raised up to one level.
What are the main characteristics of MIS?
1) Information focus. You are not focusing on data, you are focusing on information, and hence
aimed at middle managers.
2) Structured information flow. In TPS, it was structured data flow. In this case, it is structured
information, what is flowing in this one.
3) Integration of EDP or TPS jobs by business functions.
What do you mean by business function? One example is production MIS, another one is
marketing MIS, you can talk about it as personal MIS, etc. So, what we are talking here is by
different aspects, different functions of the business, whether it is production, marketing,
finance, HR, etc.
4) Inquiry and report generation. You are capable of inquiry, or putting a question and generate
relevant report based on it, usually with a database. So, multiple related relevant data is
collected together to create a database, and this database from which you are able to inquire or
ask questions, and get a relevant report or information out of the database.
So, the indention here is, you obtain a new level of information from the data that is stored to
serve management needs. So, you may want to do something as part of this, but was still built
upon information flows and data files. So, you still using information flow and data available.
For example, in a management stuff, as you grow up in the main ladder, so we give this
triangle.

So, there are three levels of management :lower level, middle level, and the top level. You have
more people at the lower level, and lesser in the middle, and very few are the top level.
So, the lower level is the one who basically do the work. The middle level is the one who make
the daily operational decisions. Other top management typically makes it. So, let us take an
example to show how this works. So, let us say there is a big super store. Let us say, this is like
200 meters long, and 100 meters wide, big superstore, let us assume it this way.

Child section is far apart from the alcohol section, just to make sure that the children does not
go to alcohol and do not drink or something like that. This store keeps electronic data of all
sales across the country. So, you have all electronic data, every sale data is logged. Then let us
say, you analyse the sales data and find that the people who tend to buy diapers in the afternoon,
also tend to buy beer. So, maybe they get tired by changing their kids diaper, we do not know.
So, when you found that the people who buy diapers in the afternoon, also tend to buy beer.
That is the information you have. This, you did not know earlier. So, then it does not make
sense to keep them this far. You may want to move them to the same level. So that the distance
is reduced. Even though it is separated a little bit, but the people can fetch alcohol for quickly.
Now, once you have this, this movement of alcohol section from this extreme corner to nearby
the kids section is an operational decision made by the store manager, and once a decision is
made, is a decision good or bad, how do you know? Then, you collect more sales data, after
relocating, and see whether sales increased.
If the sales have increased, after you moved the alcohol department from one corner to near
the child section, and you get more money out of it. So, as from the daily operational aspect to
make profit, it is a good and wise business decision.