AN EXPERT SYSTEM SHELL FOR MIS CONCEPT S E U I N G

ABHINAV AGGARWAL PROJECT MANAGER, CMC LTD. 1 EASTERN AV, MAHARANI BAGH,NEW DELHI

ABSTRACT :

All investment decisions must be based on sound analysis of the possible outcome of the implementation. A simulation model to help us quantify the benefits accruing out of an Information System Implementation as per specific environment of the user is discussed in the paper.

1. INTRODUCTION

A structured and balanced Management Information System (MIS) is as vital for an organisation as the nervous system is for human beings. But at the same time its design is equally complex. It requires a balanced compromise between the capabilities and requirements of an organisation.An organisation may require an extensive MIS with networking facilities and sophisticated hardware I software, but may not afford it financially.This expert system leads the user to determine an appropriate compromise for the extent of installing an MIS keeping in view the financial constraints of the organisation. It also displays the savings that wouM be incurred by installing an MIS in the organisation.

We assume here that the functions whose information flow is to be automated have been identified already through a scientific "feasibility study".

2. CONCEPT-SELLING

The philosophy of "concept - se1ling"is inherent in several marketing aspects but a precise scientific study has been done only recently 111. Selling an idea or a concept to an interested party through a thorough scientific analysis is what the concept - "sel1ing"intends to do.The benefits and the utilities of the concept are highlighted in clear tangible terms to convey the advantage of adhering to the concept.The concept is projected as a compromise between the requirements and the bonstraints of the interested party .So a concept is sold in realistic and logical way.

3. MIS

The concept under consideration here is Management Information System. Management,as Stanley Vance saysjs the art of decision'-making.A manager relies heavily on the provided information for making a decision. This information concems the organisation's objectives, policies and current status. Any discrepency in the information may cause great disasters and crisis.So managing this information is crucial for the efficiency and survival of the organisation.

KULBHUSHAN MALHOTRA

ANEESH KHETARPAL B. Tech. REC KURUKSHETRA

ti

A management information system is the one that may refer to one or more computersAn MIS may be used at various levels in the organisation ranging from the clerical personnel(record keeping) to the higher management(deasion making}. Generally, computers are used for data processing and also as aids for decision making.

It should be emphasized here that the concept of MIS can,in no way,be generalized for any two organisations. Every organisation has its distinct environment. Thus design of MIS is specific to an organisation depending upon its functions,resources, policies,needs and constraints. This expert system does not attempt to give a generalized solution Rather it tries to capture user's environment first and then optimizes the cost of installing MIS for that organisation. That's why our first endeavour is to delve in the user's specific environment to optimise the cost of MIS.

3.1 TWO PLATFORMS : USER OR DESIGNER?

The software has two options-firstjor an ordinary user who may use it as an expert system for optimizing the cost of installing MIS in the organisation and the second for a designer who may use it as a "shell" to develop higher own rule bases.

3.1.1 USER

The first option(user) does not require any technical expertise on part of the user except forfamiliarity with the organisation and its rule bases are constructed on statistical analysis and are inflexible.

3.1.2 DEESIGNER

The second option(designer)is expected to be used by a person who is well versed with the trends of the industry.A rather statistical knowledge of various factors affecting the industry would help the designer to modify the rulebases according to the specific needs of the industry. This option (designer) provides full flexibility for modification with detailed description of every "why"."what" and "how" of the expert system.

The rule bases of option one have been tested on some experimental data and were found out to be valid for the specific environments.Still better accuracy can be achieved by designing one's own rule-bases at the designer's platform.

3.1.3 WHY DESIGN ?

The rule bases for user's platform are rigid and inflexible.These have been constructed by very general thumb rules and can not be modified at the user's platform. So, we offer some flexibility at the designer's platform to modify these rule bases as per specific industry's requirements.

32

It is expected that if a designer makes a design file using the designer's plalform for a particular industry, the users of the same industry can conveniently use this data file while using the expert system. Otherwise, the default values would be taken.

Here we assume that the extent to which a particular factor affects the organisation is same for all the organisations in the industry. Clearly, then, the design files are "industry-specific "while the results of the expert system are"organisation-specific".

3.2 USER'S PLATFORM :UNDERLYING PHILOSOPHY OF THE EXPERT SYSTEM

This expert system gradually captures the environment of user's organisation through an interactive session. The user is asked a few general questions like no. of employees, turnover of the company,its age,no of remote sites etc( which can be answered comfortably by any user )to capture the environment of the organisation.8ased on these responses a parameter " Urii-size" is calculated from the rule bases.This parameter is in proportion to the size of the organisation which is a measure of organisation's capability to invest. It goes on further to ask for the various functions(at several remote sites) which the organisation requires to be automated.their relative importance and networking requirements.The cost of installing MIS in user's environment is calculated through predefined rule bases. If this cost does not suit user's capability to invest , an option is given to go back to the specification of requirementsJhereby changing some of them and optimising the cost. After suitable RerationsJhe user comes out with a balanced compromise between the organisation's capabilities and requirements.

A further session determines the payback period and the number of person hours saved which may encourage the user to subscribe to the concept of MIS.

3.3 DESIGNING THE EXPERT SYSTEM

The cost of MIS for a large organisation is normally more than that for a small organisation.Our approach is to calculate the cost in terms of "units" for a specific structure and then to determine the "unit-size" for the organisation This upon multiplication gives the net cost.

3.3.1 DETERMINING THE UNIT-SIZE

We assume that following factors primarily determine the size of an organisation.Their relative importance is approximated both by the designer (by rule-base) as well as by the user(by assigning a priority).

(a) number of employees (b) turnover (c) number of piant-sites (d) age of the organisation (e) collaborations with foreign companies

Our rule for designing the rule-bases has been-"Higher the value of any of the above factors,higher is its contribution in determining the size".But the user may not agree far a specific environment and may assign low priority to a factor to which a high priority has been assigned in the rule-base. The unit-size is calculated as a compromise of the two values(one entered by the user and the other assigned by the designer). In the expert system, user's priority is decreased deliberately in comparison with the designer's value to redlice any intuitive errors on part of the user. This is because we expect a designer to have a better knowledge and understanding of the environment than the user.

3.3.2 STEPS IN DESIGN

We proceed as follows. (i) The user is asked to enter the no-of-employees in the organisation. A corresponding value U1 (to be used later) is obtained from the rule-base(table 1 ,given at the end of this document). (ii) The user is asked to give a priority P1 on a scale of 1 to 9 (1 - least imp, 9-most imp) to the number of employees in determing environment. (iii) This priority is converted to a lesser value (so as to enhance designer's contribution in determining the final result and to reduce intuitive errors on part of the user) as per the rule base table 5(b). (iv) Similarly.the no-of-plant-sites and the turnover-in -last-fiscal-year are asked and U2 and U3 are obtained from rule-bases 2 and 3. (v) The user is asked to give priorities P2 and P3 to above factors and the converted priorities are obtained from rule-base 5W.

(vi) The user is asked whether the organisation has any foreign collaborations. If yes,U4 = 1 else U4=0. (vii) Priority P4 for the foreign collaborations is asked and converted as per rule-base 5(b). (viii) If the age of the organisation is more than 15 years, U5 = 1 else U5 = 0. (ix) Priority P5 for the age of the organisation is asked and converted as per rule-base 5(b). (x) Unitcount is calculated as--

Unitcount = (U1 * Pl)+(U2 * P2)+(U3 * P3)+(U4 P4)+ (U5 P5)

(xi)This is mapped in rule-base 4 to obtain the unit-size of the organisation .This factor i s in proportion to the size of the organisation . We are to see now the type of management required in an organisationh different environments,different functions may be used4 is not possible for a designer to determine precisely the different functions which are there in a specific environment(0f hislher end user).So we present an estimated list of various functions out of which the user is asked to select and assign priorities to the functions used in hisVIer environment.

As individuals, all of us are different from each other.Our perceptions and decisions are biased by our past experiences.ln this context it is interesting to look at the following definition :

"An organisation is not a mere conglomeration of certain functions and processes. It is ,in fact, a seething caldronof emotion, perspiration, nobility, foolishness, greed, sincerity, selfishness, idealism, vanity, and generosity (all of which are issues related to intangible aspects).lt is far uglier sight than a beehive or an anthill,and it is far more difficult to comprehendlts actual operation is almost impossible to diagram because it is shrouded in a fog raised by the heat of human activity."[]

To approach precision in our calculations, we need to convert intangible parameters to tangible ones .We attempt to do this making specific queries about user's requirements and reducing intuitive errors. We make a general module for calculating the cost of automating various functions. This would be used repeatedly in the further analysis.

33

3.4 FUNCTIONS

A MODULE FOR DETERMINING COST OF MIS FOR

An estimated list of various functions is displayed to the user.The organisation may not be. using all the functions or the designer may be interested in adding some more functions to the list .The option of adding more functions is given only to the designer and not to user.He/She is asked to assign priorities to various functions on the scale of 1 to 9(l=least imp. 9=most imp). An 0 is entered i f the function is not relevant in the organisation. Firstly, the complete list is shown with proper elaboration of the terms.The priorities are accepted and mapped into the rulebase (table 5(a)) to calculate no.of units of each function. Then the cost of installing MIS for these units is calculated by the following expression:

functioncost = noof-units' unit-size

(Total cost of all functions=sum of individual functioncost) where noof-units = Sizeconstant convertedgriority (Size Constant i s a constant which varies for different organisations and is assigned by the designer )

- 4. LIST-OF-FUNCTIONS

1 .PRODUCTION :production planning and control,product- design etc. 2. promotion, advertising, pricing, packaging, transportation.

3. PERSONNELIHRD: training, recruitment, career- planning and development etc.

4. CORPORATE PLANNING: (a)strategic p1anning:long term planning. (b)tactical planning: short term planning. (c)operations consistent with strategic and tactical planning.

5. FINANCE and ACCOUNTING: granting credit, collection processes,cash management, loans, leasing, sales of stock, budgets, cost-analysis, recordkeeping etc.

6. CUSTOMER ORIENTED FUNCTIONS: consultancy, maintenance, billing, customer support, warranty, hand-holding etc.

7. SECURITY: watchmanship, insurances, security of secret data, security of organisation, virus protection etc.

8. DlSASTERlCRlSlS MANAGEMENT: combating disasters and several hazard situations.

MARKETING AND SALES: marketing, sales, product-

9. ENTERPRISE RESOURCE PLANNING: planning of all re- sources for the enterpriseJnfrastructure etc. This module returns the cost of automaton of the functions cited by the user in priority.

- 5. --- TYPEOFMIS:

The user is aqked to select one out of following : (a) CENTRALISED SYSTEM: In this, there is a single MIS for controlling all the sites and functions of the organisations. (b) DISTRIBUTED SYSTEM: In this type of system, a standalone MIS is there for each site without any centralised Control. (c) ENTERPRISE WIDE NETWORK: This is a distributed system with a centralized MIS at one of the sites. We have to deal with each case separately.

34

(a) In case of centralized system,we proceed as follows: (1) Execute the function-cost module.The output we get is the cost of automating MIS for the functions at centralized site. The various functions are managed from this site only. Clearly,the information has to flow from various sites to centralized site . (2) Ask the user whether a computer network is required at the central site.lf yes,calculate networkcost=Ob functioncost Totalcost-functiomst + networkcost (b) In case of distributed system,we proceed as follows: (1) Ask the no.4-sites from the user. (2) For first site,execute function-cost module.This gives the cod of automating functions at that site. (3) Ask the user whether a similar MIS can be used at other sites also i.e.whether there are more sites similar to this(in functions). (4) If yes, get noof-similar-sites from the user.

(5) Cost of MIS at these similar sites is given by noof-similar-sites function-cost (6) Our next step is to determine the remaining no of sites which is given by (no-of-sites)-(noof-similar-sites). (7) Now,steps(P)to(6) are repeated until there are no more sites. (8) We now calculate the total cost by summing up the individual costs obtained in step(5) (9) Ask the user whether a local computer network is required at these sites. (10) If yes,then network-cost is added to the total-cost obtained so far where networkcost=0.3 Total-cost (11) If no,then the step(l0) is skipped. (12) Ask the user whether a computer network to join the isolated islands of automation is required. (13) If yes,Totalcost=l.l Total-cost (14) If no,then skip step(l3)

If no, noof-similar-sites=l

Thus the total cost of installing MIS in distributed system is determined. (c) In case of an ENTERPRISE WIDE NETWORK ,there is an independent standalone MIS at each plant site and the overall control is governed by an MIS at centralized site. The cost of standalone MIS at each site can be calculated as done in case(b)i.e.in distributed systems. Here the cost of network will also include the cost of networking centralized site to each of other sites.The cost of centralized MIS can be calculated as the sum of following: (1) The function-cost at the centralized site which is calculated by functioncost module. (2) The cost of management at central site. Clearly, this is a function of the no.4-sites because all these sites are controtled by central site.So, cost of mgmk0.01 no.of-sites total-cost so far. Thus ,depending on user's choice whether helshe is having centralized, distributed or enterprise wide network, we display the total cost of installing MIS in the user's environment with the details regarding how many functionslunits have been automated and whether they have a computer network or not and to what extent. The user may not be able to afford this cost and may wish to vary the "extent" of installing MIS in hislher organisation. So,he/she is given an option to trace back the whole process to optimize an affordable cost. Ultimately,the user comes out with an optimised cost which is a compromise between higher requirements and capabilities to invest.

6. MIS BENEFITS

Now, the user has to be told about the benefits which would be accruing i f MIS is instatled in the organisation. A major contribution to this would be the reduction in various hold-ups and delays that may be caused otherwise. So, here, we are going to calculate the payback resulting from cutting down these delays by MIS.The cut down in delays can be broadly discussed in following systems: 1. Non-MIS vs MIS systems 2. 3.

It is presumed that all organisations do have a structure of information flows.In view of this,we assume that atleast a non- automated MIS is a must for the existence of an organisation.& such,in whatever form it may exist,MIS is a vital organ of the 0rganisation.However.a complete study of existing procedures and practices in the company only can bring out the effectiveness of the existing MIS system in the organisation. As carrying out such a study is outside the scope of this paper, we are not quantifying the improvement in response time in the context of a non-MIS vs MIS oriented orgn. This should not in any way undermine the importance of following a stluctured MIS even if it is only a manual one, would definitely improve the overall effectiveness as well as cut down on the delays in any organisation. The cut down in delays actually means the no. of person hours saved. Although this is not tangible as such but we are trying to make it tangible by determining its dependencies on certain factors. The no. of person hours saved depends on a no. of factors.We proceed as follows to find the person hours saved: (1) Get the no.of-functions from the user whose information was manually managed initially and now it is automated. Greater the no of such functions,greater will be the no. of person hours saved. (2) Now,map the no-of-functions onto a rule base (table6) and find corresponding value of V1 from table. (3) Get the no.-of-sites from the user.If the no. of sites are large and are non-networked,then inter site transaction causes large delays. (4) If no-of-sites> 1, then ask user, no of sites which are connected(or networked). (5) Map the no. of sites connected onto rule base(tab7) and find corresponding value of V2 from table7. (6) Get the no-of-trasactions/week(of the functions automated) from the user.Greater the no-of-transactions,greater will be the person hours saved. (7) Map the no-of-transactions on to a rule base(tab8) and find corresponding value of V3 from table& (8) The three factors(discussed above in(1),(3),(6)) are not equally responsible for finding person hours saved.So,the relative importance of these factors is reflected in weights as given in next step. (9)

We have assigned weights of 2,1.1.5 to indicate the relative importance of these factors. (These constants can be changed by the designer) (10) Obtain the person hours saved from table 9 (rulebase) corresponding to the value of Vcount. (11) Aftecfinding the person hours saved,the next step is to determine the payback. Pay-back(in rupees)=no-of-person hours-saved cost of one person hour. First factor has already been determined and the second factor can be calculated as:

CALCULATION OF PAYBACK : QUANTIFICATION OF

Manual MIS vs Automated MIS Networked MIS vs non-networked MIS

We have to calculate Vcount as: Vcount=2 V1 + V2 + 1.5 V3

(12) Cost of one person hour depends upon the size of organisation which is reflected in Ucount.Thus we can modify our formula by putting cosVperson hour = Ucount i.e. our formula becomes Pay-back(in rupees)=no-of- person hours-saved * Ucount. Thus, payback is clearly reflected and hence cut down in delays is made tangible to the user in terms of payback.

DESIGNER'S PLATFORM

DESIGN FILES

We have provided a default design file which can be used at the user's platform.lf required ,a design file specific to user's needs can be created at the designer's platform.

EXTENT OF FLEXIBILITY

The rulebases to calculate Ul,U2 .... etc. are fully flexible and can be given ANY values.lt is pertinent on the part of the designer to verify that the values filled by him\her do not yield absurd results.However,guidelines are provided for making the rulebases and warnings (not ERRORS) are displayed i f the designer attempts to assign too large or too small values.

CONCLUSION :

This expert system may be used as a primary tool by the users who are planning to revamp or install MIS in their organisations.The rulebases can be updated by making new design files.At best it can provide reasonable results depending on the validity of the design file it uses. The system helps those planning an investment into MIS to evaluate the effectiveness of their decision in terms of pay back period and person hours saved. Such analysis can help in quantification of the benefits accruing out of an investment into IT (Information Technology).

REFERENCES

[ l ] Aggarwal Abhinav, Wadhwa S, 'A concept selling algorithm for CIM'. proceedings of TENCON 93, IEEE Region 10 Conference, Beijing, China.

Teicholz Eric, Orr Joel N, 'CIM Handbook' Mc Graw Hill Designing with system series, USA 1987.

[2]

T h e v a r i o u s t a b l e s ( r u 1 e b a s e s ) u s e d i n t h e

D E E ' A U L T d e s i g n f i l e are s h o w n a s fo~lows:

TABLE 1

I ' NO-OF-EMPLOYEES

1..100 101. .200 201. .400 401. . 600 601. . 800 801.. 1000 1001. .1500 1501. .2000 2001. .3000 3001..4000 4001..5000 5001. .10000 > 1 0 0 0 0

35

I

1 - -. - - I I I I I I I I I I I I I I

U1 _ _ _ _ 1 2 3 4 5 6 7 8 9 1 0 1; 1 2 13

TABLE 2

-1- - - - - - - - -

- 1 - - - - - - - - -

- - - - _ _ _ _ _ _ NO-OF-SITES I U2

1. .2 I 1 3. .4 I 2 5. .6 I 3 7. .8 I 4 9. .10 I 5 11. .12 I 6 13. .14 I 7 15. .16 I 8 > 16 I 9

- - - - _ _ _ _ _ _

TABLE 5(a)

TABLE 3

I

- 1 -

TURN-OVER (RS.CRORE) 1 U3

< 0.50 I 1 0.51. .1.0 I 2 2. .3 I 3 4. .6 I 4 7. .10 I 5

16. .20 I 7 21. .30 I 8 31. .50 I 9 51.. 100 I 10 101. - 2 0 0 I 11 201. .500 I : 12

- - _ - _ - _ _ _ _

11. .15 I 6

501.. 1000 I 13 1001..5000 I 14

>5000 I 15

TABLE 4

- 1 - - - - - - - - - - - - - - _ _ - _ _

Ucount I U-size I

3. .20 21. .40 41. .60 61..80 81.. 100 101.. 120 121.. 140 141.. 160 161.. 180

>180

I 5000 I

I 9000 I 13000 I 17000 I 20000 I 23000 I 25000 I 26000 I 27000 I 28000

-

36

- I - - - - - - - - - - - - - - - - - - -

I C O W . PRIORITY PRIORITY

0 1 2 3 4 5 6 7 8 9

I 0 I 1 I 1.25 I 1.50 I 1.75 I 1.95 I 2.15 I 2.20 I 2.25 1 2.30

TABLE 5 (b)

- I - - - - - - - - - - - - - - - - - - - PRIORITY I COW. PRIORITY

0 1 2 3 4 5 6 7 8 9

I 0.0 I 1.0 I 1.8 I 2.5 I 3.0 I 3.4 I 3.7 I 4.0 I 4.2 I 4.3 I

TABLE 6 - - - - - - - - - - - - - - - - - _ _ _ _

I NO-OF-FUNCTIONS I v1

- 1 - - - - - - - - - - - _ _ _ _ _ _ _ _

0 I 0 1 I 10 2 I 17 3 I 25 4 I 30 5 I 32 6 I 34 7 I 36 8 I 37 9 I 38

10 I 39 > 10 I 40

I

TABLE 7

- I - - - - - - - - -

- 1 - - - - - - - - -

_ _ _ _ _ _ _ _ _ - CONNECTED-SITES I v2

2 I 5 3 I 10 4 I 15 5 I 20 6 I 25 7 I 30 8 I 35 9 I 40 10 I 43

> 10 I 45

_ _ _ _ _ _ _ _ _ -

I

TABLE 8 _ - - _ _ _ _ _ _ _ - NO OF TRANSAC./WEEK I v3

1 I 3

3 I 15 . 4 I 20

5 I 25 6 I 30 7 I 35 8 I 40 9 I 45

10 I 50 > 10 I 60

I

2 I 10

TABLE 9

I - - - - _ - - - - _ - - - _ _ _ _ _ _ _ _ _

Vcount I MAN HOURS SAVED/DAY - - - _ - - _ _ _ - _ - - - - _ - _ _ _ _

I 15. .30 I 100 31. -45 I 200 . 46. . 6 0 I 300 61. -75 I 400 76. . 9 0 I 500 91.. 105 I 600

1 0 6 . . 110 I 700 111.. 125 I 800 126. .140 I 900 141.. 155 I 1000 156. . 170 I 1100 171.. 185 I 1200 186. .200 I 1300 201. .225 I 1400

I

37