YT YouTube2Text

YouTube2Text → Transcripts → Lecture 45 : Demand Forecasting

Lecture 45 : Demand Forecasting — Transcript

by NPTEL IIT Kharagpur · 6,016 words · 884 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:01[Music]
0:22[Music]
0:26hello dear friends welcome back to nptl
0:28online course on Logistics and Supply
0:30Chain management so in the last session
0:33we uh discussed about locating the
0:35facility so where we should locate the
0:38facility so that we we can be somewhere
0:41we can be uh you know cost optimization
0:45can also be ensured and somewhere the
0:47responsiveness quick delivery of the
0:49product can also be ured right so uh
0:53once you have decided the location the
0:56very next step should come the
0:59forecasting the demand because once you
1:01will predict the Demand only then you
1:03can design the capacity of that location
1:06right how large how big you need to
1:08design so this session we'll discuss
1:11about the demand forecasting right what
1:13are the factors we need to consider what
1:16are the challenges for demand
1:17forecasting what different types of
1:19demand forecasting techniques we have
1:21we'll discuss some qualitative
1:23techniques we'll discuss some
1:24quantitative technique as well right so
1:28this is uh some related data latest data
1:31by World Bank on uh Indian you know how
1:35Indian economy is behaving uh since last
1:382 three years right so if you can see uh
1:42instead of you know uh the global
1:45economy uh the global markets they are
1:47facing challenges but still India
1:50somehow managed uh the growth rate right
1:53and you can see Financial year 2023 2022
1:57and 3 uh at 7 2% right so India's growth
2:02rate was the second highest among G20
2:06countries and almost twice the average
2:09for emerging market economies right so
2:12this resilience was underpinned by
2:15robust domestic demand which is quite
2:18obvious we have used customer potential
2:20whatever we are producing we can you
2:23know consume within the country right so
2:26that that kind of Market is there and
2:29strong public infrastructure investment
2:32private investment is also coming and
2:34then if you see Bank credit growth rate
2:37has also increased to
2:4015.8% uh from 203 24 to
2:4313.3% right so uh from last year
2:4713.3% so if you talk about India's GDP
2:51so that is also predicted somewhere more
2:53than 6% right so uh if you talk about
2:58manufacturing sector is a tremendous
3:01growth we are seeing Services sector
3:03again matured at that maybe we are
3:05somewhere struggling with the
3:07agriculture sector right so because of
3:10maybe the growth rate is not that high
3:13what we were expecting right if you talk
3:17about the expenditure approach method so
3:20uh GDP growth in the third quarter was
3:22aided by a strong uptake uh in private
3:26investment spending which I was talking
3:27about that private investment is also
3:30increasing and
3:3110.6% year over year we have seen the
3:34growth which is not small right so
3:36talking about the private consumption
3:38improved to
3:403.5% and uh uh this again coming back to
3:44India's resilient domestic demand which
3:47is you know uh not only a opportunity
3:51for the local players but also then the
3:54global players now they are thinking of
3:56investing uh within India and uh the
4:00kind of the business environment now we
4:04are providing here in India with the you
4:06know one uh Point teex structure and all
4:10those liberal policies are there so kind
4:12of red carpet welcome is there for all
4:14the foreign investors so you can see uh
4:18the growth is also coming in that way
4:20right so let us come back to some
4:23production and supply chain perspective
4:25if you see gross value added is uh is
4:29growing at 6.5% year-over year so that
4:32means that is showing the growth in
4:34manufacturing which is again you can see
4:37and construction activities are also
4:39increasing at
4:4199.5% and then you can see Services
4:44sector somewhere 7 to 8% we have seen
4:47right agriculture which I was talking
4:49about the rate is not that high but the
4:53main point here is that gap between GDP
4:56and gva but gross value adding we can
4:58see it's not only one year problem year
5:01over year we have seen that there is GDP
5:04is higher than the uh gross Val uh value
5:08added right so now this reason what is
5:12the basic reason so very clear
5:14indication that demand is high right but
5:18there is gap between the demand and
5:20Supply we are not able to produce all
5:22the things we need here that's why we
5:25are going for uh importing right so if
5:28you talk about food industry only so
5:30food industry if agriculture growth is
5:32Contracting at 8% you just imagine so
5:36where from where we will you know
5:38produce that M so that Gap is obvious
5:41Supply Gap is obvious right so that's
5:43why we need to predict our demand in a
5:46proper way so that all the facilities
5:49can be you know uh made ready to
5:52whatever future challenges will come
5:55right so demand forecasting is the
5:57process of understanding and predict
5:59meting the customer Demand right and why
6:02we are doing that when you obviously how
6:06we are meeting the demand through that
6:08long supply chain in that we need to
6:11manage our suppliers also we need to
6:13place the replenishment orders with them
6:16right whatever demand is accordingly
6:18they will supply the material
6:20accordingly we'll store the material in
6:22the storage house accordingly we'll
6:24design our manufacturing capacity
6:26accordingly we'll design our
6:27distribution Network so that means all
6:30depends upon only demand forecasting
6:33having said that that means we need to
6:35define the demand function very
6:37clearly that means demand function on
6:41which the demand is depending maybe
6:44maybe the economic condition maybe the
6:46price of the product maybe the price of
6:49the competitive product maybe the
6:51inflation rate maybe the so many other
6:55factors right price of the complimentary
6:58Goods so many other factors can be there
7:00so you need to configure this demand
7:03function very carefully for your
7:06industry for your product or Services
7:09whatever you are selling so once you
7:12have done that only then we can ensure
7:14that this is not happening not we are
7:17producing over excess production is not
7:19there and we are also not stock out
7:22right in the market so obviously this is
7:25not uh one step we have one
7:28comprehensive structure how we can
7:30predict the Demand right so what are the
7:34factors we'll just go through what are
7:35the factors which are affecting the
7:37demand seasonality which is very obvious
7:39now the summer season is going on right
7:42everywhere there is demand for AC's
7:44water coolers for uh fans all those
7:49things are going on right so that is
7:51seasonal but what about after 3 4 months
7:54then there will be winter season right
7:56so then how you need to predict it very
8:00accurately because the industry I was
8:03talking about maybe when there is off
8:06season during Winters you can still you
8:08know pile up your inventory and whenever
8:10the season will come you can consume
8:13that inventory right that is one way but
8:15again if it is a kind of perishable item
8:17how you will maintain the
8:19inventory right so seasonally
8:21seasonality plays very important role
8:24right and suddenly if Peak demand is
8:27there can you go for expanding Your
8:29Capacity like anything to meet that Peak
8:32demand and then after 3 months what you
8:35will do with those extra capacity those
8:38extra resources Manpower machines
8:41whatever you plant you have expanded
8:44simply you cannot fire the people you
8:46have hired now for your Peak demand and
8:48then after two months you cannot do that
8:50right then competition this also plays
8:53important role while shaping your demand
8:55function suddenly some player are losing
8:58Maybe
9:00uh market share because of their brand
9:02image like what happened with Nokia
9:04Nokia was leader once uh in a time in
9:07Indian market and you can see because of
9:10their after sale Services people were
9:12relying in Nokia right but then the
9:14issues happened with B battery and then
9:17you see now where is Nokia so if that
9:20situation happened with Nokia that
9:23becomes opportunity for other players so
9:25then either your demand can go down or
9:28skyro
9:29that is obvious if Zo is coming with so
9:32many different offers obviously demand
9:34for other H bodone ceron Network all
9:38that has gone down so then they
9:41collaborated to fight still they could
9:43not right so competition is also
9:46important factor geography so geography
9:49from where the demand is coming where
9:51are your customers and where are your
9:53manufacturers right so uh let's say if
9:57I'm talking about some hilly area and
9:58you are prom in your low floor car
10:01whether it will work in hilly area low
10:03floor car no obviously so then you know
10:06that that is specific for some
10:09particular reason if you talk about the
10:12Cosmetic industry the cream they are
10:14selling right so hilly area weather
10:16conditions are different uh here in
10:20kakur or Kolkata weather conditions are
10:22different so the demand will be you know
10:25if we are using all those stuff right so
10:29that geographical location also you need
10:32to consider
10:33economy so you you might have seen
10:36during recession there is usually fall
10:39down in the economic activities during
10:41uh 2008 recession also we have seen
10:44during this covid pandemic we have also
10:46seen that demand for automobile industry
10:48demand for all those luxury product uh
10:52have gone down right but the demand for
10:55low price product like basic things
10:58right people start buying like anything
11:00like your grocery item so demand is
11:03increasing for those item price will
11:05shoot up like anything but luxury items
11:08price may fall but even then you are not
11:09interested because you are not getting
11:11enough income you want to survive so for
11:14survival you need those minimum things
11:17right for whom maybe those things for
11:19getting those things you are paying
11:21extra now types of goods like I talked
11:24about the perishable Goods you need to
11:27predict in that way but if you are
11:28talking about let's say uh how many
11:31people will recharge your you know go
11:34for your data services you know that how
11:36many people are using your network inter
11:39uh this connection right so those many
11:41fixed kind of demand if even if you talk
11:43about the automobile industry once the
11:46product is matured you know this is
11:47going to be the Demand right and you
11:49know when you are going to be you know
11:51going to phase out that product and you
11:53are coming up with the new model and
11:55then you are shifting in a proper way
11:58right so that transition is planned
12:00transition but what about the perishable
12:02items so then you need to predict
12:04properly advantages of uh demand
12:07forecasting obviously you can plan your
12:10inventory in that way so that you are
12:12not maintaining extra inventory that
12:14extra carrying cost you can prevent
12:17holding cost you can prevent and the
12:20extra effort required to maintain that
12:21inventory and then you don't want to be
12:24stock out so for that you need to
12:25predict your demand accurately cost
12:28reduction when you can reduce inventory
12:30obviously you can reduce the cost as
12:32well not only inventory cost but the
12:35production cost but the transportation
12:37cost extra you are producing extra you
12:39are transporting extra you are
12:40Distributing to your retailer and then
12:42your retailer is struggling to consume
12:45that in the market because there is not
12:47no demand customer satisfaction so once
12:51you predict the demand accurately you
12:53can be easily available in the market
12:56with your product and services and you
12:58can satisfy the customer and in that way
13:00you can maintain loyal customers
13:03production planning you can go as per
13:05your Demand right and resource
13:07allocation that maybe next 2 three
13:09months peak season is there I can
13:11prepare for that with extra resources
13:14right or maybe right now I can plan
13:17inventory little bit and then I can meet
13:19the demand so whatever way you want to
13:21work and different strategies are for
13:23different kind of
13:25Industries marketing and promotion now
13:28you need you know that that the demand
13:29for particular product right so you know
13:33which product you need to campaign more
13:35and which you can restrict with the
13:37campaign if you're phasing out any
13:39product no demand is coming after your
13:42heavily investment in the marketing
13:44things but even then you are struggling
13:45with you know arranging that so then you
13:49should you know you can plan that
13:51marketing budget as well financial
13:53planning how much you need to invest
13:55into raw material how how much you need
13:58to invest into
13:59uh capacity planning production
14:02warehouses distribution so that planning
14:04you can do right risk mitigation if you
14:07are doing the forecasting thing
14:09regularly if any fluctuation is coming
14:12in the market that you can assist in
14:15advance and then you can mitigate that
14:18risk right strategic decision making how
14:21much I need to make another strategic
14:23decision where I need to locate the
14:25warehouse Center if that demand is
14:26coming high demand is coming from kak
14:29let us set up one Distribution Center in
14:31kakur right so and new product
14:34development Market expansion capacity
14:36expansion all that will depend upon
14:39obviously you will you will go for
14:41capacity expansion when you are sensing
14:43that your demand function is increasing
14:45when you are getting that input supply
14:47chain optimization you can do based on
14:49that once you are sure about the future
14:53then you can optimize and seasonal
14:55planning any peak season scale up scale
14:58down you can do and if you are doing the
15:00demand forecasting in a proper way you
15:03have that in uh uh information and you
15:05are taking informed decision obviously
15:07will provide you the Competitive Edge
15:09over the other players what are the
15:11challenges of uh uh demand forecasting
15:15obviously all how we are predicting the
15:18demand future based on information so
15:21that means always you need information
15:23data to process to find out the future
15:26Demand right so if you are lacking with
15:29the data or sometimes we are lacking
15:31with the resources to get that much data
15:34then we are uh facing difficulty to
15:37predict the accurate demand and because
15:40how we are predicting the demand because
15:41we are basically what we are doing in
15:44forcasting we are taking the historical
15:46data we are trying to extrapolate the
15:49trend for future we can see that how it
15:51will impact if the demand is behaving
15:54somewhere like this so we can see it is
15:56growing so we can extrapolate that this
15:59will grow like this only at this rate or
16:01maybe like this it will come down
16:03depending upon whatever factors you are
16:06considering right limited visibility
16:08into future always you cannot predict
16:11the future accurately so some risk is
16:14always there and in forecasting risk is
16:16always there right so technological
16:19change will come and you never know that
16:21this technology will be outdated
16:23whatever you are doing right now and you
16:26need to change to the new technology and
16:28you have buil up all all the in
16:29inventory production you are doing so
16:32much inventories with the retailers with
16:34the warehouses how you will consume if
16:37better technology product is available
16:39in the market who will go for the uh
16:41outdated product inaccurate demand
16:43models if you are not predicting which
16:45model is required for predicting the
16:47demand that also you need to pick
16:49carefully right so we have qualitative
16:51quantitative models we'll see how that
16:53will help you to you know accurately
16:55predict the demand poor forecasting
16:58process see this forting forecasting
17:01process is depending upon collecting
17:03data so in during collection also
17:05whether you are touching all the data
17:07points or you are collecting few samples
17:10then that process you can just imagine
17:14how if you are collecting only 500
17:16samples or if you are collecting 5,000
17:18or 50,000 right so there will be
17:21difference between the results when you
17:22are compiling the
17:23results insufficient resources
17:26collecting 50,000 data points you need
17:29those resources you need technology you
17:31need to invest whether we have efficient
17:34those many resources or not and
17:36sometimes the factors like pestel
17:38analysis we are doing political
17:40economical social legal ethical
17:43technological that all those external
17:46factors you cannot predict right so if
17:48any change is there so that will change
17:51type of forecast demand forecasting
17:54passive demand forecasting is a kind of
17:56uh where you are not considering the
17:58Dynamic environment and whatever is the
18:01demand in the last year you are thinking
18:03that similar kind of demand will be
18:05there so you are using though for this
18:08kind of model passive demand forecasting
18:10where steady demand is there active
18:14demand forecasting for any startups or
18:17new companies coming with new products
18:19new Ventures new you know R&D you are
18:22doing with the old product so then you
18:24need to predict accurately you need to
18:26consider the active envir en Dynamic
18:29environment and then that is active
18:31demand forecasting means quickly you
18:34need to update your forecasting
18:37short-term demand depending upon the
18:39time maybe that shortterm you are saying
18:42that somewhere less than one year
18:44whatever you are
18:45predicting 1 year to 2 year maybe the
18:49medium
18:50term forecasting or more than 2 years
18:54again can be your long-term
19:01forecasting but that again depends upon
19:04this this is again shortterm medium-term
19:07long-term depends upon the nature of the
19:09industry this we discussed earlier as
19:11well if you're talking about
19:12entertainment industry launching web
19:14series or movie today you are launching
19:17tomorrow maybe it is gone if you have
19:19seen once no one is going unless it is
19:21you know that level so that means for
19:25that one or two days or initial one or
19:28two weeks are important as shortterm
19:30planning medium-term maybe after that
19:33how you launch on different platform
19:35longterm there is no longterm if you're
19:37talking about fashion industry so
19:38shortterm you need to predict accurately
19:41so that you can plan just in time right
19:43when it is required so day-to-day
19:46routine activity You can predict and
19:48long-term demand forecasting so over La
19:51next maybe five years how you are
19:53thinking if you need to expand the plan
19:56plant capacity whether you have that
19:58land available with you so that you can
20:00expand right so overall blueprint maybe
20:03for 10 years 20 years targeting your
20:05vision statement how you are looking
20:07yourself after 20 years right so macro
20:11and micro level micro level within that
20:13industry how other players are competing
20:16how your product is competing with the
20:18other products available in the market
20:20all those factors related to product
20:22compliment products right the prices of
20:25the raw material all those things you
20:27will consider macro like I talked about
20:30the pest analysis always political
20:32economic social technological ethical
20:35legal all those factors you will
20:37consider in macroeconomics so talking
20:40about demand forecasting techniques we
20:42have qualitative techniques as well we
20:44have quantitative techniques so quickly
20:46we'll go through each technique and
20:48we'll see first is dely Method so
20:52different steps are there but I'll in
20:55nutshell I'll tell you dely method is we
20:58we have group of 10 to 15 experts right
21:02they all are sitting around round table
21:05and then we are giving them one
21:06questionnaire or common agenda to
21:08discuss right so we are asking to record
21:11their inputs they are recording their
21:13inputs we are collecting those inputs we
21:15are compiling those inputs and then
21:18again we are circulating inputs of all
21:21the experts among themselves right so
21:26after that going through Those comp that
21:29compiled report now they will come to
21:31know okay we missed these points and
21:33someone else maybe they have mentioned
21:36those points so then they will start
21:38thinking on that we'll do this exercise
21:40again we'll compile again we'll
21:42circulate until the consensus is done
21:46right so this is complete dely method so
21:49uh you can just go through step by step
21:51selection of expert 10 to 50 experts as
21:54I told you the expert in that industry
21:57where you if you talk about automobile
21:59industry so 10 to 50 Experts of
22:02automobile industry two wheer industry
22:04or four-wheeler industry whatever you
22:06are going to predict they will sit
22:08together right then questionnaire design
22:10what are the points you will include in
22:12the questionnaire on which you are
22:15seeking the input right those
22:17quantitative
22:18qualitative uh questions close-ended
22:21questions can be there open-ended
22:22questions can be there as I mentioned
22:25first round of questioner you will
22:27provide your inde independent inputs but
22:30then because once you are providing your
22:33independent input you are not biased
22:36with other inputs what others are
22:38thinking so first I want to record your
22:41independent opinion on that then we'll
22:44take the feedback analysis from each one
22:47we'll summarize that and subsequent
22:49record Rec rounds will happen where I
22:52will circulate the inputs coming from
22:55all 50 experts to all the 50 experts and
23:00then again we'll do this exercise and
23:03we'll final report we'll prepare where
23:05all points are now 50 experts have given
23:08their points all points will circulate
23:11then they will come with the common
23:12agenda that this is going to be the
23:14forecast for the next 6 months next 5
23:17years next 10 years in that particular
23:20industry right then this is another
23:23method Salesforce composite forecast
23:25usually this is we are doing when we are
23:28you know dealing with those products
23:29which we are consuming through retail
23:31industry usually so uh either you talk
23:35about uh the fmcg products or if you
23:38talk about uh uh other food industry or
23:41even not only food construction industry
23:43also right where you are talking about
23:45moving uh cement or steel roads or
23:48bricks whatever required uh In the End
23:51Market so again here uh what we are
23:54doing uh we have sales this is also
23:57known as pulled Salesforce composite
24:00method why pulled Salesforce composite
24:03method because a reason wise we have
24:06defined that this reason particular
24:08markets 10 Market this person sales
24:10person is looking after maybe he is
24:13visiting those retailers 10 markets
24:15maybe you have 20 25 retailers he is
24:18managing data for all those retailers
24:20how much daily inventory they are
24:21depleting how much inventory they are
24:23depleting of the competitors what is the
24:26market demand how it is going to behave
24:28they are interacting with the retailer
24:30getting the fresh inputs directly from
24:33the customers as well so right now that
24:3610 Market information he will be sharing
24:39maybe at the next level District level
24:42so this is maybe region small region is
24:44there then District level I'm collecting
24:48from maybe 50 these sellers are there in
24:52one District then that District
24:55different district one state we will be
24:58comp comp iling the data from one state
25:00we will be compiling data for one
25:02country from one country we'll compile
25:04data for one continent so this is how if
25:07you are my MNC this is how you will
25:09compile data pooled Salesforce map so
25:13data collection again step by you will
25:16uh uh see where you can collect the data
25:19obviously with the retailers choose
25:21forecasting method how you will predict
25:22how you'll compile the report how you'll
25:25generate the initial forecast and then
25:27collaborative input as I told you reason
25:29wise then District wise will be compiled
25:32then state wise will be compiled then
25:33country wise will be compiled and in the
25:35end we'll find out one figure for that
25:38particular product in India how much
25:41demand is going to be right so another
25:45is expert knowledge here what we are
25:48doing expert we are conducting
25:51interviews with the expert and we have
25:54listed question and we are giving them
25:57either close-ended question or
25:58open-ended questions or this different
26:01discussion points this is also known as
26:03focused group discussion or dely method
26:06can also be included here or
26:08brainstorming sessions can also be
26:09included here and then we'll ask them
26:12their opinion perspectives right on the
26:15production of uh that particular
26:18industry right so how it is going to it
26:20is if it is talking about we are any
26:22technological development will be there
26:24or not for next 5 years so how they they
26:27are expert in that field they can uh
26:30tell us that this technological you know
26:33advancement is going to happen and we
26:35need to move in this way so this is
26:37another method survey method like uh uh
26:41this can be rooted through your
26:42Salesforce or maybe special survey team
26:45can be there who will be defining the
26:47objective of the survey obviously in
26:50this case we are predicting demand so we
26:52will be interested in knowing the demand
26:55of the product for the whole industry
26:57for our product right then we'll design
26:59the questionnaire and multiple choice
27:02questions like I talked about liquid
27:04scale question ranking all those
27:06open-ended questions can be there and
27:08then you will select the samples like
27:10different markets consumers you will go
27:12customers you will interact with and
27:15then you will collect the uh survey
27:18right so then again you will Analyze
27:20That survey report you will interpretate
27:22those results you will tabulate results
27:24you will come up with the diagrams that
27:26how it is going to be right so again we
27:29can predict Demand by using the survey
27:31method so either you can take online
27:34survey like monkey surveys is there
27:36Google form you can telephonically you
27:38can collect the information face tof
27:40face survey also when you are
27:42interacting directly with the customer
27:44usually when we launch new products then
27:46we are going for face tof face
27:48interaction or you can send some emails
27:51and then you can collect the information
27:53this I talked about interview or focused
27:55group discussion Where You Are you know
27:58interacting with the experts and getting
28:00that expertise uh information from them
28:04now let's talk about some of the
28:06quantitative methods so quantitative
28:08methods means now we'll record the uh
28:12inputs in terms of quantities or we'll
28:14see historical data we'll try to figure
28:18out the you know pattern in the
28:21historical data let's say if this is how
28:23the historical data is moving
28:29so I can say this is the kind of product
28:32which is where demand is cyclic means
28:35there is cycle so peak season is there
28:38then off season is there then again peak
28:40season or maybe the demand is like this
28:43if it is like this so I can see it is
28:46increasing like this right so this how I
28:50can draw this I need data points
28:53quantitative data so then I can exploit
28:55the trend if I'm having up to this that
28:58I have so I can see after this it is
29:00going to be like this so I can predict
29:02for future right so first method is
29:04moving average where we are considering
29:09uh latest data sometimes we are
29:11providing the weightage also to latest
29:13months data and then we are finding out
29:16what is going to be the demand for the
29:18next month we have three methods for
29:20first is simple moving average second is
29:22Ved moving average third one is
29:24exponential moving average right simple
29:27moving average Let's uh talk about with
29:29the some example only see I'm having
29:32data for January month February month
29:34March month let us take 3 months moving
29:37average so what I'll do I'll sum up
29:40these three 150 + 160 + 170 divided by 3
29:46this is will be this will be the average
29:48will be the demand for March month let's
29:53now take for April if I want to
29:55calculate for April I'll take the last 3
29:57months
29:58February March April if I'll calculate
30:01May I'll take now I'll not take February
30:04I'll take March April May right so
30:08latest 3 months I will consider so you
30:10can see for March it is there for April
30:12it is there for May June July you can
30:15find out latest but here what we are
30:18doing we are giving equal weightage to
30:20last 3 months that should not be because
30:23in some cases if the latest month data
30:26is showing you you know uh some
30:29fluctuation in the demand either in the
30:31positive side or negative side and then
30:33we are giving equal weightage to all the
30:35months then we can predict it wrong so
30:38that we need how we can tap that
30:40exponential moving average where we are
30:42giving you know weightage right how uh
30:46for this the basic purpose of this is
30:49that we we are we need only one month
30:52data earlier we were calculating 3
30:54months data right and then we can uh
30:58capture those fluctuations as well right
31:01so uh the unique feature is it
31:03incorporates all past data points
31:06because we are calculating the
31:08exponential moving Factor right from all
31:11the past point and then we need only one
31:13month data to predict it and if it is
31:15nonlinear if it is not moving like this
31:19right if it is moving like this so that
31:21also we can uh configure with this so
31:24how we can predict the demand for
31:26particular month t so we need to find
31:28out the uh smoothing factor which is
31:31Alpha and we can find out Alpha by 2 / n
31:35+ 1 n is number of months we were
31:38considering so earlier we were
31:39considering three so if you are taking
31:41three months you can take three if you
31:43are taking five you can take five so 1
31:46minus Alpha into your uh last month
31:50predicted demand so let's take here
31:53example so if you are taking three so
31:55obviously we can calculate the EXP
31:57exponential smoothing factor right so 2
32:00ID 3 + 1 it will be .5 and then the
32:04alpha will be 0.5 for this Alpha will be
32:060.5 and we if we are calculating for
32:10March month so then for March month it
32:12will be 0.5 * by 170 + 1
32:18-.5 multiplied by last month predicted
32:22demand let's say predicted here is 155
32:25so it will be 155 this is how you can
32:27calculate so you can see for February
32:29month you need demand actual demand for
32:32February month and January month last
32:34month predicted demand and you can see
32:37this is for each month you can
32:39calculate and this is 3 months
32:42exponential moving
32:44average right now next is baited moving
32:47average in this case what we will do we
32:50will like I was talking about uh giving
32:53highest priority to the last month so if
32:56I'm calculating for March maybe I'll
32:58give 0.5 weightage 50% weightage to Mar
33:01Munch data maybe I'll give 3% weightage
33:04to February month data and I'll give
33:07vitage to 02 to January data so how I'll
33:10do 150 multipli by 02 + 160
33:15multi3 + 170 * .5 this will be demand
33:20for March month so now you can see here
33:25different scale has been used 1 2 3
33:27before right that also you can use four
33:30means latest month you have given
33:32highest weightage this also you can use
33:34where I have given a highest weightage
33:36to uh March month now you can find out
33:41uh when I will calculate for April
33:43highest vitage will be given to the
33:45latest month right this is how you can
33:47find out the next is Trend projection so
33:52Trend projection obviously uh we will
33:55find out we'll plot all the data points
33:57his historical data and then we'll find
33:59out the best fit line right so that
34:02linear or simple linear regression
34:04multiple linear regression or polinomial
34:06regression we can use for that so we'll
34:09see how we can fit that data see uh this
34:14is like let's say this is X independent
34:18variable this by is dependent variable
34:20I'm having like these data
34:23points so I'll draw the best fit line
34:26maybe this is the best fit line which
34:27which is touching maybe passing all the
34:30points nearby we are drawing that line
34:33which is having minimum distance from
34:34the data points it can be in the
34:36negative direction as well which will
34:39reflect that your demand is decreasing
34:42as this X Factor is increasing this is
34:45positive relationship this is the
34:47relationship can be like this also
34:50relationship can be exponential also
34:53right so this trend we need to find out
34:55if I I can predict this L up to this
34:58point I can extrapolate for next month
35:00this is going to be the demand for next
35:02month this is going to be the demand I
35:03can predict for those points but I can
35:05predict only up to the maximum scale we
35:09have considered right so linear this is
35:13how we can predict linear where this is
35:15your by intercept this is your slope of
35:17the regression line by intercept means
35:20where this line will touch this a point
35:22is your by intercept and slope of this
35:25line is this angle right right so in
35:28detail simple linear regression if you
35:30can go then uh maybe you will find out
35:33how we are driving this equation of uh
35:36by intercept plus m means the slope B
35:40into your independent variable which is
35:42T here uh X here this independent
35:45variable and dependent variable is Y
35:47where here it is written dependent
35:49variable is
35:50y right so this m we can calculate this
35:54is slope right and the uh
35:58what is the significance of this slope
36:00is per unit change in this independent
36:03variable how much it is going to affect
36:06the dependent variable if it is value is
36:09.55 that means per unit change in
36:12independent variable will affect
36:150.55 times dependent variable so this is
36:20exponential uh Trend model you can use
36:22this also where we are using exponential
36:24function again we are using a e e to
36:28power b b is your growth rate and a is
36:31your initial value we can use polinomial
36:34equation as well where straight line
36:36will not be there and you can find out
36:38polinomial equation I'll just give you
36:40one example of simple linear regression
36:43where you can see this is how we can
36:45calculate see this is the simple this is
36:47the regression equation like I told you
36:51this is uh dependent variable demand
36:53this is independent variable this is
36:55your slope this is your B intercept how
36:58we can calculate the by intercept B note
37:01is equal to by bar minus B1 xar B1 how
37:05we can calculate the slope this is how
37:08we can calculate the slope we have X
37:10values we have y values right let's say
37:14this is one question I just want to
37:15predict we are launching different tv
37:18ads and we are have how many units of
37:21car we are selling so this these data
37:23points are there we want to draw this
37:25line so what should be the slope of this
37:28line right so this is use this is number
37:32of ads independent variable and your car
37:34sold depending upon number of ads by so
37:38we can calculate B1 from here you can
37:40calculate xar you can calculate by bar
37:43this you can find out and then you can
37:45find out B not as well so just you put
37:48the value here B1 V not any x value you
37:51put three it will give you the number of
37:54cars sold right so this you can
37:57extrapolate same equation you can use
37:59regression Line This is known as simple
38:01linear regression right so uh for more
38:05details uh maybe research methodology
38:08course you can see where we are using
38:10this uh Trend how we are drawing the
38:13best fit line right and then we are
38:15talking about the uh error as well right
38:18there is error component as well which I
38:20have not included because that is
38:21statistic part right so maybe in detail
38:25you can uh refer to regression line so
38:27this is all about the quantitative
38:29methods right so these are references
38:31you can refer thank you very much

About this transcript

This page contains the full transcript of Lecture 45 : Demand Forecasting by NPTEL IIT Kharagpur, generated from the public captions YouTube serves with the video. The transcript has 6,016 words across 884 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.