 Dear participants, welcome to the course of supply chain digitization. So, it is jointly being taught by Professor Piyanka Burma, Professor Sushmita Narayana, Professor Deva Brathadas from IA Mumbai. So, in today's session we will talk about analytics in supply chain management. So, this is the first lecture of module 3. So, now let us start with what is analytics. So, you must be hearing this term day in and day out and this is the buzz word in today's time. So, we need to know what are the various characteristic of analytics, what are the various components of analytics. So, the first and foremost before I do any analytics I need to have a clear objective in mind like why am I doing the analytics. So, what is the goal am I trying to achieve after doing this. So, if I have that clear cut objective in mind with that goal I should collect data. So, the first and foremost data is an very important part of analytics. So, once we collect the data there could be some discrepancy in the data. So, I have to clean it, process it. So, once I clean it, process it I get good data then from there I need to develop model. So, the model could be statistical model, it could be AIML based model, it could be optimization model. So, whatever model you develop you have to make sure that the objective what was there in your mind earlier the model should serve that purpose ok. So, then only the model will be useful. So, once we develop the model we test it validate it then the model should be useful for taking better decisions. So, there could be multiple decisions which decision maker can take, but the model would give me an analytical way to take better decisions. So, once I get the decisions then I have to implement this in the organization. Once I implement it the implementation should create value to the organization. So, now, if I summarize it in analytics I have four important characteristics. The first one is data using then second one is model, the third one is decision and fourth one is value creation. So, if I summarize in one line it would be like this. Use data to create models which will lead to decisions that create value. So, if I have to summarize analytics in one sentence this could be the sentence. Now, as you saw in the last slide data is a very important part of my analytics. So, now if we see the trend I have the data for last 15 years which I got it from Statistar.com website. In 2010 1.2 trillion gigabytes of data were created, captured, copied and consumed worldwide. In 2015 it became 15.5 trillion gigabytes. In 2020 it went up to 59 trillion gigabytes and then we had covid. So, in the covid period lot of company digitize themselves. And if I go for digitization obviously more data will be created, captured and consumed and that happen. And now if we see the prediction in 2025 the total amount of data which will be created, captured, copied and consumed worldwide will be around 175 trillion gigabytes. Now you can see in last 10, 15 years the data volume has been increasing exponentially. So, therefore, I can see that big data are being created. So, I need to know the characteristic of big data ok. So, therefore, let us move to the next slide which talks about big data. So, now if you see here there are multiple these are there the first one is volume, the second one is variety, third one is velocity, fourth one is variability, fifth one is value and sixth one is velocity. So, these are the important v of big data using this 6 big v we can actually characterize the big data. So, let us first start with volume. So, the volume it says huge amount of data that means, the large volume of data which is being created captured, consumed, copied every day across the world. So, we will take an example and explain this concept called volume and the customer order data. So, I am sure you must be purchasing items from e-commerce website like Flipkart, Amazon then nowadays Q-commerce has also come. So, or in all of this platform it if you see specifically let us take an example of e-commerce. So, where you can place an order throughout the day 24 by 7 from any place and you can send the order to any location. So, therefore, 24 by 7 the data is being created and captured in their e-commerce website. So, the data could be like first you are searching in the website like what item you like, what product you want. So, then after searching you are shortlisting. So, you have search data in the e-commerce website then you are shortlisting then you are checking out. So, after you check out then you have to select the address now in which address you want to send the products to. So, after you check out then you have to pay for it the payment could be through credit card through debit card then we have UPI nowadays also and then sometimes cash on delivery option is also there. So, all of this information are being captured in the e-commerce platform. So, imagine across the globe 24 by 7 millions of customers are using this e-commerce platform. So, amount of data which are being generated is huge voluminous data. Then we have also another example called inventory count this is very special example from like if I see from the point of view of large warehouse distribution centers and this is very important concept and lot of data are also being generated over there. So, take an example of distribution center you have let us take an example of e-commerce distribution centers. So, obviously they will have varieties of product and for each product I need to know exactly for how many units this particular product is there in their distribution center. Let us say in ERP system enterprise resource planning system the data shows 10 units, but in reality you might have 9 units there could be 1 unit which might be this misplaced or in reality it could be 11 unit also or it may not have scanned properly. So, this kind of discrepancy are very common in distribution centers. So, how to avoid this discrepancy I need to count the inventory physically and therefore, I have to make sure that whatever inventory are there in physical system it has to match with the ERP system. So, this process is called inventory count and through inventory count process lot of data are also being generated. Then the next V is variety next V is variety. So, let us see so, variety means multiple data sources and forms. So, what do you mean by multiple data sources and forms? So, this could be let us say image data this could be audio data this could be video data this could be text data and so on. So, therefore, there are varieties of data which can be created copied, consumed and used for model development purpose or decision making purpose. So, let us take an example of customer review. So, if you see customer review suppose you buy a product from somewhere then you have to give review suppose you did not like the product you can give a review let us say very bad bad poor. So, you can enter as a number sometimes they ask categorical question is it good, bad and so on. So, the option would be excellent, good, poor, very good and so on. These are the categorical data then sometimes you may also enter the description of the product or description of the problem. So, in a text format you enter whether I like the product or not if I do not like what part of the product I did not like or what part of the service I did not like. So, you enter a text. So, that is an example of text data then also many a time you give an example of image data let us say. So, suppose the product is damaged. So, I will take the image of the product and then put it in the e-commerce website. So, that is an example of image data. Then I also have audio data and nowadays if you see in Google you can actually do audio search. Suppose I want to search something instead of writing or typing I will say this word. Let us say I want to find the best book in the world. If I want to search this I will just tell I want to find the best book in the world. This is an audio kind of data which is being captured over there. Then also I have video type of data. So, what happens? I will talk about AGV. So, AGV is an example in which video data is also being generated. So, AGV means automated guided vehicle. So, if you have visited high end warehouse where AGV are there. So, they move across the aisles of warehouse. They take videos in and around and then give us the video to the user. So, this is an example of AGV of video data. Then we have an example of quality inspections. So, this is an very interesting example specifically in the context of supply chain. This is very important. So, if you have seen chip, there are lot of circuits are there like small small circuits are there and if there is any mistake in the circuit then there will be error. The chip will not function properly. So, I have to inspect each and every chip. Other is customers will complain and very small like it goes in mobile phone imagine how small it is. So, if I have to check manually it is very strenuous for the aisles and sometimes we make mistake human make mistake. So, therefore, there will be error. Sometimes the good quality chip might be treated as bad quality. Sometimes bad quality chip might be passed as good quality. So, both the things are problematic for us. So, therefore, what we need to do? We need to check properly and make sure that the proper quality chip is going to the market. So, nowadays lot of technology has come to do this and one of the analytics which is being used for quality inspection is called image analytics. So, what they are doing specifically in chip industry nowadays? So, they have the example of best quality chip which is super circuit is fine. So, that is benchmark image and that is tested against each and every chip which is passing through the manufacturing facility. So, I will take the image photo of each and every chip. Then we will compare it with the benchmark chip which is the best quality chip. If there is any discrepancy then automatically red flag will be highlighted and I will know that this chip has problem in the circuit. So, not only the discrepancy will be highlighted, it will also tell like at exactly at which point or which location of the circuit there is a mistake. So, engineer will get to know that this part of the circuit there is a mistake. So, that is how image analytics are being used heavily nowadays. Then I will also give another example of video analytics. So, we now talked about image analytics. So, I will also give an example of video analytics. So, if you like go and visit some of the best warehouse in the world European Union USA. So, now slowly this technology is coming as we talked about inventory count as we talked about inventory count in the previous example. So, in warehouse I have to make sure that whatever number of sq's are there in my warehouse it has to match in the system. If system says 10 physical warehouse also should say there is 10 products in the warehouse. So, this has to match. If there is a discrepancy then there will be problem. What would be the problem? Like if I say I have 10 units sq in my warehouse and in reality I have only having 9 then I will not be able to serve my 10th customers ok. And if I say 10, but in reality if I have 11 then one product is unaccounted. So, which will lie here and there and obviously it will become obsolete. So, I have to make sure that whatever number is shown in the ERP system the same number should be shown in the physical warehouse also. So, then how do I check it? If you go to a large warehouse like thousands of sq sometimes lakhs of sq's are there which are lying in the aisles and most of the aisles are like some horizontal some vertical very difficult to check each and every day and its manual work. So, it will take lot of time. Now, video through video analytics through video analytics companies are able to do it. So, what do they do after the factory shutdown after the warehouse is shut down in the night? They put the video on top of UAV unmanned aerial vehicle that is drone. So, drone is fitted with the video camera and then drone moves across the aisles. They scan the QR code of each and every aisle products and in the morning report is being generated. So, if there is a discrepancy of particular sq it will be shown over there. So, manual inventory count is actually been replaced by video analytics. So, it is very good example and how companies are also using this technology. So, these are about video. So, there are many such example. So, now let us move to the next we call velocity. So, what is velocity? The data generated at a very high speed real time. So, real time data and it is being generated at a high speed. So, every fraction of second lot of data being generated. We talked about this in e-commerce. So, in e-commerce 24 by 7 you can place an order from any part of the world. So, lot of being data are being generated at a high speed and real time. So, I will give another example of GPS tracking of vehicle. So, nowadays like good logistics service providers they have trucks in which GPS is installed and it is very like it is not so costly technology to adapt. So, companies are adopting it also. So, they have GPS installed in their truck. So, whenever truck is moving from one location to another location I would get to know where the truck is exactly located. So, this is very important because let us consider that you are sending material from Mumbai to Guwahati. So, long distance from west to east we have to travel 3-4 days of travel and if I am sending a high value item. So, basically I am more worried. So, I need to know where the exact location is. So, if I get to know the GPS latitude and longitude of each and every second then I will know where the truck is and let us say GPS location is fixed for next 1 hour, 2 hour then I will know that the truck is waiting over there. So, I will be able to track it where it is currently located. I will also be able to know if there is any problem or problem or not. So, let us say one truck has been ideal the GPS location is same for the next 5-6 hours. So, obviously, I will be alarmed I will know that there might be an issue I will talk to the driver and get it sorted. So, this kind of real time data will obviously be useful for me and definitely logistics facility will be more efficient. Then we have another example of real time data and nowadays across the world is also being implemented is monitoring drivers eyeball because mostly the drivers are driving in the night this long distance big trailer trucks and driving in the night because during the day traffic will be there and city entry and closure are like closed. So, therefore, mostly they prefer in the night and obviously, if you drive in the night you are prone to fall sleep. So, if driver falls asleep obviously, accident will happen. So, now through image analytics through image analytics now through image analytics what companies are doing they are tracking the capturing the image of the eyeballs like now every seconds and these image are being sent to the server. So, if I can see that eyeballs are not moving drivers eyeball are not moving for the 2, 3 minutes let us say continuously. So, obviously, driver may fall sleep. So, therefore, I will put an alarm if there is no movement in eyeball for like so, continuously for some time and then the drivers will be alarmed you will get up press and again start driving. So, that is how we can also minimize the accident. Then we have a next 4 V which is variability. So, what do you mean by variability? So, I have like variability within a particular kind of data. So, there is a difference between variability and variety. So, in variety I have variety of data forms it could be image, audio, video, text, but within a particular kind of data let us say it is a text data. So, within a text data set I will have variability. So, we will give an example of demand data. So, now if I take the example of demand data for an for electronics items ok. So, during Diwali in India obviously, the sales goes up then again the sales come down. So, there are variability in the debt demand. So, sometimes demand is low, sometimes demand is high. So, this is called variability. And now because of the digitization social media and customers behavior are changing a lot. So, therefore, variability has become the part and parcel of today's data. So, in 20, 30 years back the data was not so variable. Obviously, there were seasonality, there were trends, but variability has increased nowadays exponentially. So, how do I capture this variability that is a challenge. And specifically if you see some of the industry like in which customers behavior plays a very significant role, the data has become much variable. So, whatever items we used to like last year customers used to buy in huge volume, the same items are not being sold. Specifically I will give an example of beauty products in FMCG domain. And it is directly linked with customer behavior, how customer perceives this product and so on. So, therefore, the lot of variability are there. So, if I see next last 2, 3 years data for maybe for particular brand particular item customers were liking it. And now the same products are not being sold in the market. The may be the new competitors have come and they have taken away my demand. So, therefore, whatever advertisement you are seeing in the media, social media, what topics are being discussed, this has direct impact on my demand variability. So, these 4 like volume, variety, velocity and variability. These are 4 main pillars of big data. In addition to this we have 2 important V, the next one is value. That means, whatever data is being generated, it should be useful for my organization or it should be useful for me, it should be useful for my team, then only it makes sense. Then the next one is veracity. This is very, very important topic called veracity. So, that means, the data generated should be accurate and come from a reliable or trustworthy sources. So, one good thing is digitization happened and social media are there, Twitter, Instagram. So, we are getting news updates fraction of seconds. So, information is plenty available to us and it is fraction of second in one part of the world, whatever is happening I will get to know from another part of the world. This is fantastic, but what happens this also creates problems in trust of the reliability or trustworthiness of the data. So, therefore, I have to make sure that whatever data I am getting it should be accurate and it should not be a fake data, it should not be fake image or it should not be a wrong data. So, if I enter wrong data obviously, my model also will give me the wrong output. So, therefore, as a data scientist I have to make sure that these 6 v are being monitored properly. I have to check what is the volume of the data, what is the variety of the data, what is the velocity, is there any variability in the data so on and whether it is bringing value to the organization or not and lastly the veracity should be there. That means, data should be accurate. So, these 6 v characterize my big data and these are very very important like going forward like as a participant or as a student, you have to make sure that you follow this big v 6 v's then you will not have any like difficulty in understanding the characteristic of big data. Now, we understood the concept of big data so in the next class we will talk about like analytics. So, like different kind of analytics which we can do using this data and develop models which can help in supply chain domain. So, with this we stop this video. So, thank you all the participants look forward to seeing you in the next lecture in which we talk about analytics and types of analytics and its application in supply chain management domain. So, thank you look forward to see you in the next video.