YT YouTube2Text

YouTube2Text → Transcripts → Lecture 20: Project Estimation Techniques (Contd.)

Lecture 20: Project Estimation Techniques (Contd.) — Transcript

by NPTEL IIT Kharagpur · 4,134 words · 603 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00[Music]
0:17okay good afternoon
0:18now let's start in this section in this
0:22lecture we will take up some of the
0:24examples for this and rest a function
0:28point methods will take few examples so
0:30that you can better understand we'll
0:33take some more examples on this al Bai G
0:36function points then we'll take two more
0:38types of concern points that's C months
0:40about two function point and the cosmic
0:42function points so we have already
0:45discussed these parametric models for
0:47resides last class I have told will use
0:50I will see portrait type shop for these
0:55estimations like IP IEP function pants
0:59malfunction pants cosmic function pants
1:01and Kakuma two one in kokomo two last
1:03class already have discussed IEP
1:05function points now we'll see this
1:07Simmons mark to concern point and cosmic
1:10function points before going to that
1:11let's take one or two more problems more
1:14examples on this IEP G function points
1:18so we will take up this example but I
1:21hope that before going to this example
1:23the let me again tell you the let us
1:28take up that a this example and here
1:32we'll see that how this previous formula
1:37that we can use I have already told you
1:39last class first we have to find out the
1:41unadjusted function point then we have
1:44to refine it by using the value
1:47adjustment factor so this formula for on
1:50adjusted function point can be written
1:53as what I have already shown it in the
1:55last slide this will be called do
1:57summation of number of elements of the
2:08given type may be DJ input or output or
2:15query what is the number of elements
2:17into the weight the weight I have
2:23already given you four simple there can
2:26be three
2:27the weights can be divided into three
2:29categories depending on whether it is
2:30simple or medium or high
2:33so that table algorithm ulta that
2:36complexity classifiers the multipliers
2:40we have already given you in the last
2:42class and now with this now let us take
2:45up this problem so now let us take this
2:49one more example this example said that
2:53consider a project with the following
2:55functional units that are suppose in
2:57that project there are 50 user inputs 40
3:00user outputs 30 pipe user enquiries and
3:036 user files for external interfaces so
3:07now let us assume that all the
3:09complexity adjustment factors and
3:11weighting factors they are average
3:12everything is average so then the
3:14question is that compute the function
3:16points for the project and then compute
3:20what will the sides
3:21given that the program needs 70 LOC per
3:24function point so now let's go you can
3:28see that the inputs is equal to 50
3:32outputs is you can do 40 and courage is
3:3635 and internal files you can see that a
3:39number of internal files is 6 and 4 with
3:44the external interfaces that 6 and this
3:48so use these values in the given formula
3:51that I already have given you and then
3:54it will give you this value of ufp so
3:57basically two things I am taking the
3:59possible number of elements from each
4:03category input-output inquiry then
4:06internal file and external interfaces
4:09multiplied by the multiplier the weight
4:13which is whether it will vary depending
4:15on whether it is simple or average or
4:17what hi in my previous problem I have
4:20already told you that they are all what
4:22average so all the complexity addressed
4:25are just meant factors are abolished as
4:27well as DC weighting factors they are
4:29operates it will be much more easy to
4:31compute so now you have P is equal to
4:33strict photo perform formula apply this
4:35you will get 628 then you have to find
4:38out the what value are just one factor
4:40again that formula we have told you 0.6
4:42a plus 0.65 plus 0.01 into TDI we are a
4:47TDI is equal to total degree of
4:50influence and here put in parameters
4:52since it is told that this is average
4:54simplicity I have already told you that
4:55these complexity of different factors
4:58are average as you like the weighting
5:00factors them in the GS see this general
5:03system Catholics there they are these
5:06are also average so again for average
5:07since you have taking a scale from zero
5:09to five again this is average you will
5:11be coming three so this will go and
5:13point zero seven so the AP will be more
5:15the adjusted function points will be
5:18equal to on other state function points
5:20in to be AF so because why we are doing
5:25this we want to refine it so the you can
5:28get the adjustment function point by
5:30what multiplying the unadjusted function
5:33point on the value adjustment factor so
5:35this is given two six seven two so this
5:37is they are just add function points so
5:41now what is question again
5:43and the second part question is that
5:44find out the size of the complete
5:46project how can compute if I find out
5:48the complete the size of the complete
5:50project you have to use this information
5:52it said that the program needs 70 LOC
5:56part function point that means per
5:57function point that can be seventy lines
5:59of code so what is the total number of
6:01lines of code so sorry how did the total
6:04number of AP function point 672 so the
6:07total side u will be equal to the
6:09function points into LOC parkinson point
6:12so 672 into seventy in this way you will
6:15get that these total size of the project
6:19in terms of LOC is coming to be 474 zero
6:23so in this way for small projects you
6:26can easily use these formula find out
6:29the function points and sites so in the
6:32examination some this types of small
6:34what questions might be asked and you
6:36have to find out the answer so let's
6:38take quickly another example in this
6:40example a project you have to develop
6:44for that you have to find out the
6:46function point for that project this
6:49project has the following
6:50characteristics there are number of user
6:52inputs is 30
6:54number of future outputs is sixty number
6:56of you did incur is twenty port number
6:58of piles is equal to eight and number of
7:00external interfaces to so and assume
7:02that all the complexity are just going
7:04to value their average and all the
7:06weighting factors are updates so then
7:07you have to find out the number of
7:09function pairs you can see your P
7:11against Pawar formula that I have given
7:13summation up on the number of elements
7:16from its type into weight whole then
7:19take the overall summation you will get
7:21applying this formula straight forward
7:23you can see that there are 32 number up
7:25for what input users and do we know
7:28therefore average case this what value
7:31it support it is already given these are
7:34just reactors are average so the value
7:36will be what 32 into port similarly
7:38output is 60 the average weight is 5 and
7:41this is the enquiries it is 24 number of
7:44encourage are there you can see and this
7:47average value for inquiry is for unlike
7:51this you can compute there are eight
7:53numbers of what internal files and two
7:56number of external interfaces so 8 into
7:58average value for the interfaces 710 and
8:01to number of external interfaces average
8:04value it's seven so find out this value
8:06it's coming 618 then you have to compute
8:10the value adjustment factor by using the
8:12formula 0.6 here PI plus 0.01 into TDI
8:15and since it's a given that these are
8:18also these general ket system
8:20characteristics they are average so I
8:21can put it 14 into 3 so that will 1.07
8:25so finally the unadjusted function point
8:27can be obtained by multiplying the
8:29unadjusted function point and the value
8:32adjustment factor so this is giving to
8:34be 6 6 1 point 2 6 so the total number
8:36of are just add function points or the
8:38adjusted function point counts for this
8:40project will be equal giving us 606 one
8:43point 2 6 so suppose if a poor function
8:47point you have to write say 80 lines of
8:49code so what will the size in terms of
8:51LOC so 661 1 to 1618 so this will give
8:55you roughly the size of the proposed
8:57project so in this way you can use all
8:59breast function pointer method to find
9:02out the to estimate the size of a given
9:05project so now let us quickly say about
9:08these what function points mark 2 this
9:12is also another interesting approach to
9:15find out the concerned parents this was
9:17developed by Charles R Simmons actually
9:20the details of each method is available
9:23in a book developed it has published a
9:26book on suppress IDing and estimating by
9:28EULA and John's in 1991 which contains
9:31the details of his function pairs that
9:33is function points mark 2 it is built on
9:36actually an extension up for this al
9:39breast function point so it's build on
9:41work by Al just one simple method it is
9:44developed in parallel to I opinion app
9:47it's so simultaneously it was developed
9:50in parallel with the AL breast I appear
9:53pits so it is rather simple and more
9:57readable method so how it works
10:00so for each transaction you have to
10:03count the data items input the data
10:05items output and the entity types assess
10:09so these three parameters you have to
10:12count and then you can find out the
10:16total function point for the whole
10:18project by using a formula which we'll
10:19see in the next slide
10:21see how does it work sighs I already
10:23told you you have to supply the input
10:25items agile as you will see also three
10:29parameters that you could have told the
10:31number of input output the number of
10:33input out items supplied to the process
10:35the number of entities assessed from
10:37this data store and the number of output
10:40items produced so this in this way this
10:43function point mark to this works this
10:46is a simple is a simpler method than
10:48this al based on some point and it is
10:52widely used in UK so what is the final
10:55formula for finding out the mark to
10:58function point the function count is
11:01given by the following equation and I
11:04start zero point five eight plus any
11:06star one point six six plus I know star
11:080.26 we are ni any and I know we have
11:11already defined earlier and I did at
11:14items input you know the data items
11:15output and any the NT drives associated
11:18associated or the sorry the entire
11:21assess so this value judge do these
11:25constants are some weights zero point
11:27five eight one point six six and zero
11:28point two silhouettes this Simpson okay
11:33see months he had found out these values
11:38of the constants by performing
11:39experiments in his lab so directly he
11:41had utilized these values of the
11:44constants along with these parameters ni
11:46any and they know and he has developed
11:48this formula to find out the function
11:50points this is known as Simmons mark two
11:53function points this is simpler than
11:54alastor function point now let's see
11:58about the whether these function points
12:02so far we have seen the algorithm and
12:04some points on the Simmons function
12:05points can they be extended to real-time
12:08systems and embedded systems see this
12:11mark - and IEP purity function points
12:14they were mainly designed for
12:16information systems under hence they are
12:18not suitable for embedded systems why
12:20because these two function points they
12:23are largely or heavily dominated by the
12:27input and output operations the most
12:29emphasis is given on the number of input
12:31operations on the output operations so
12:34in order to overcome this problem so
12:36cosmic full point cosmic pull function
12:40points they have been developed in order
12:42come these problems this cosmic full
12:45function points they attempt to extend
12:48the concept this concept which concept
12:51this the mark to concept or iph concept
12:54to embedded systems so now these
12:57concepts have been extended to handle
12:59embedded systems to apply on embedded
13:02systems or real-time systems you know
13:03that in an embedded system it's
13:07creatures like hidden and etc okay isn't
13:11it if in an embedded system its features
13:13normally will be hidden because the
13:16software the these software's use your
13:18probably is not a human being it will be
13:21used by some hardware device or another
13:24sub third component in order to what
13:28overcome these problems
13:30cosmics deals with this by decomposing
13:33this
13:34architecture into a hierarchy of subtile
13:37layers and the embedded software is seen
13:40as being in one of body or in a
13:42particular layer in the system this
13:44layer where are this software is present
13:47this communicates with other layers I
13:49mean upon layers and the below layer and
13:51also other components are the same level
13:53same level the software component which
13:56has to be sized it can receive requests
13:59for services from layers above and it
14:02can request services from those below it
14:04this is shown in the in this figure so
14:07the software component represents
14:08suppose here it may receive what
14:11messages from higher layers it may make
14:14a request for a service what from a
14:16lower layer and also it may receive a
14:19what service from a what a lower layer
14:22it may service it will provide some
14:24service to some higher layer besides
14:26that this software component may
14:28communicate with the persistent storage
14:32pure component and other components
14:34present in the same layer this is the
14:36layered software and the based on this
14:38concept this embedded systems work so
14:40since the address the function point and
14:43assumptions upon humans function points
14:45they cannot handle this embedded systems
14:48that's why this another what function
14:52point technology has been developed that
14:54will cosmic slate now let's see how
14:55cosmics does work so so now I have a
15:01little do that here the software
15:04component that has to be sized
15:06can receive requests for services from
15:08layers above and it can equal services
15:11from those below it so this identifies
15:15what will the boundary of the software
15:17component that has to be assessed and
15:19thus the points at which it receives
15:21inputs and transmits out ports it inputs
15:25here the inputs and outputs are
15:29aggregated into data groups where each
15:32group will bring together data items
15:34that relate to the same object of
15:37interest let's see what can be these
15:39what groups the data groups can move
15:42about in four ways as follows and like
15:46one each entry and
15:48Egypt and reefs another's rights so
15:51these are the what data groups which are
15:54the four ways in which the data groups
15:56can move about so interest means
15:58movement of data into this after
16:00component that is under consideration
16:02from a higher layer or a pure component
16:04age it means movement of data out of the
16:07software component that a general
16:08consideration really means data movement
16:11from persistent storage and writes means
16:14data moment
16:15to persistent storage so in this way
16:17before which the data groups can move
16:19and what we have to do that each counts
16:22so each of the above each counts one
16:25cosmic functional side unit and insert
16:28we call at CFS you so in this way the
16:31cosmic function points can be a
16:34calculated in this to the cosmic
16:35function points can be computed can be
16:38estimated for any embedded system or
16:42real-time system the overall pull
16:47function point count how how it can be
16:50found out the overall pull function
16:51point count can be derived by simply
16:54adding up the counts for each of the
16:56four types of data moment that we have
16:58seen in the last slide the cosmic ffh
17:02they have been incorporated into ISO
17:04standard we have already seen that the
17:06LOC etc they are not included
17:10incorporated in any ISO standard but
17:12this cosmic FFPs they have been
17:14incorporated into ISO standard now let
17:17us quickly see about the what
17:19disadvantages of the cosmic a page it
17:22does not take into account any
17:23processing of the data groups okay so it
17:27does not take it does not consider any
17:29processing of the data groups once they
17:31have been moved into the software
17:32component so once they what data groups
17:36they have been moved into the software
17:38component this metric does not take care
17:40of this this material does not take into
17:41account any processing of these data
17:43groups normally it is not recommended
17:47for use in systems which involve complex
17:51mathematical algorithms so this this is
17:54not recommended for this system so these
17:56are important drawbacks of cosmic
17:58function points so now let us
18:02what are the pros and cons of
18:04difunctional parents on the positive
18:07side you can say that function point
18:08each language independent we have
18:10already seen LOC a language dependent
18:12but function points along an independent
18:14understandable by the client because she
18:17are no technical details are there you
18:19just what are the functionalities and
18:21they are associated things you are just
18:23analyzing so it can be understandable by
18:25clients which are non-technical people
18:27it is a very simple modeling technique
18:29hard to purge and the variable features
18:33decrypt and on the flip side you will
18:35see that it is very much labor-intensive
18:38to compute the what function points
18:41extensive training is required in
18:43experience may result in inaccuracy so
18:46the the function points you have
18:50computed if you are quite inexperienced
18:52it may result in inaccurate number of
18:54function players waited to file
18:56manipulation and the transactions there
18:58may be errors which are introduced by a
19:01single person so multiple raters are
19:04advised to if you are just keeping only
19:06one person to estimate this some of
19:08their sometimes that errors may be
19:10introduced so you have to put a multiple
19:12what raters to rate these what
19:15applications they can find out the
19:19different ways so multiple networks are
19:21required it does not consider one of the
19:23major problem with function point is
19:25that it does not consider algorithm
19:28complexity of a function so because you
19:30know that it rates that if there are
19:33some functions every function is handled
19:36in Sumeria but you know some of the
19:37functions are more complex you have to
19:39put more report some of the functions
19:41are very straight forward straight
19:42forward functions so they require only
19:45what less input or less time so it does
19:49not consider algorithm complexity of
19:50function so it traits all the functions
19:53similarly so I have a literal you that
19:57function point matrix suffers from a
19:58major drawback that means the size of a
20:00function is considered to be independent
20:02of its complexity which is not true the
20:04complexity of the functions are
20:06different in order to overcome this
20:08problem an extension to the function
20:10pointer metric is there which is known
20:12as future point metric so this pager
20:15bonda metric
20:16teksu knock out an extra parameter it
20:18considers an extra parameter that is
20:20known as algorithmic complexity so this
20:24parameter this parameter algorithm
20:26complexity it ensures that the computed
20:29size using the future point may take it
20:31reflects the following fact what if the
20:33fact that higher the complexity of a
20:35function
20:36the greater the important occurred to
20:37develop a if the complexity of function
20:39is high so you have to put more report
20:42to develop it if the complexity of a
20:44function is very less just like a small
20:46you a kind of a function it remain quite
20:49less support therefore it should have
20:51larger size compared to a simpler
20:53function so if we what the function is
20:57having higher complexity if you have
20:59larger sides as compared to a simpler
21:02function it should have larger function
21:04point and hence it should have larger
21:06stage as compared to a simpler function
21:08so finally let's see what the
21:11relationship or how that is the
21:14relationship between function points of
21:16a so Lucy or how a given a function
21:18point how it can be converted to a
21:20solution see for different programming
21:22languages this conversion rate is
21:25different like for C language 100 Port
21:29st.lucie is there per one function plane
21:33so one function point may contain 100
21:36for a solution similarly for C++ or the
21:39isola support function point is 53 and
21:42like this you can see it it has to in
21:44case of RC and you can see that it is
21:47almost lowest in case of HTML and visual
21:51basic it is 32 okay so how you can still
21:56make your estimation more accurate how
21:59can do accurate size estimation so here
22:02one thing you have to extract that is
22:04the risk factors because in the previous
22:06things almost we have neglected the next
22:10factors so in order to make your size
22:13estimates are more accurate you have to
22:14take the expected and then you have to
22:17multiply so you will take the
22:19requirement then estimate the project
22:21size then multiplied with the project
22:23complexity are just main factors then
22:25multiplied with the X factors so this
22:27will so these two
22:29I have already told you this is given
22:31we're in the function point analysis the
22:33project sites and they project
22:36complexity adjustment this you can say
22:39the unadjusted function point and these
22:41are the complexity factors by
22:43multiplying this you can get some value
22:47and that will multiplied by the X Factor
22:48so these two factors can be obtained
22:50from function burden analysis then by
22:54what under these two factors from the
22:56definition and the project completion
22:58adjustment and respecters they are
23:00dependent the situation you have to
23:01adjust the situation finally you can if
23:04you'll use all those things then by
23:07taking the multiplication you can
23:08estimate what the what the cost of the
23:12import and you can perform the
23:14scheduling finally let's quickly say
23:17about the object points
23:18see I don't compute objects points they
23:20have nothing to do with the object
23:22oriented programming here only the
23:25number of object points is estimated the
23:28number of object points is estimated
23:30based on three factors first the number
23:32of separate screens they are displayed
23:34then the number of reports that are
23:37produced and the number of modules they
23:39are present in the code by taking into
23:41account then by using some formula you
23:44can find out the object points these
23:45object points are normally are the
23:48they're usually simpler to estimate and
23:50that they take into account the
23:52graphical user interfaces so this is a
23:55little bit fundamental about object wise
23:57so finally we have seen that we have
24:00solved two more problems on our breast
24:04refugee function points then we have
24:07discussed the Simmons mark two function
24:10points and the cosmic function points we
24:15have seen that how cosmic point it is
24:18already included in this what ISO
24:21standards also we have discussed or to
24:25the concept of future point because we
24:26have seen that function point a Hajj one
24:29important a drawback that it does not
24:32take into account to the algorithmic
24:34complexity if two different functions
24:36are they are it treats them as equally
24:39what in fact the two function though
24:42they are there one
24:43be very highly complex another might be
24:46very less complex so which one is highly
24:48complex we have to put more rapport and
24:50who generates less complex we have to
24:52put less support so these what
24:54complexity is not taken into account by
24:56function points so that second point has
24:59been extended to in consider this
25:03complexity it has it has another
25:06parameter College algorithmic complexity
25:09so this algorithm complexity takes into
25:12account that the if the complexity of a
25:14function is more more reports will be
25:17given if the complexity Hoppe a bro
25:20function is less so less you photons to
25:22be given so if the complexity of a
25:24function is more obviously it should
25:26contain it should produce or it should
25:29have more number of concern points and
25:31hence the size will be more and if the
25:34concern is having less complexity then
25:37it should have less from some points and
25:40hence less size also we have discussed
25:42little bit a of deep object points which
25:45takes two non-count or three factors
25:47that were all three parameters okay
25:49object once it takes into three right it
25:52is based on three parameters that also
25:54we have discussed so these are the
25:56things that we have discussed on on
25:59today so in the next class we'll see and
26:02that goku model we have taken from this
26:05books we reference mainly what okay it's
26:09both of the books have been used for
26:11this these topics and finally we stop
26:14here thank you very much

About this transcript

This page contains the full transcript of Lecture 20: Project Estimation Techniques (Contd.) by NPTEL IIT Kharagpur, generated from the public captions YouTube serves with the video. The transcript has 4,134 words across 603 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.