Getting value out of data science and artificial intelligence | London Business School

[Music] good morning everyone it's a great pleasure to get to spend the next 50 minutes talking about data science with you my name is Nico Sava in at the London Business School for 12 years and in a sense I describe myself as a data scientist who's found a home in a business school so twelve years ago when I first started here data science was a niche topic we offered to pretty much zero electives on it very little teaching and it was called statistics and fast forward 12 years we have a number of data science electives that range from sort of you know machine learning and data mining to sort of you know analytics that involve simulation optimization and so forth and it's one of the fastest areas of growth for the school and our electives are very well subscribed so we're venturing into offering sort of similar courses for more senior executives in executive education and in a sense my 50 minutes today is a high-level overview of what data science can do so I'm gonna keep it more on the conceptual level I'd on the actual to the Vino methodology level I'm gonna spend maybe 10 minutes trying to show you how you do data science rather than talk about this and I'm gonna give you a little data science assignment so if you thought that you were just gonna listen and not do anything this morning after a case so here's my little data science assignment for you so it's based on the game of Russian roulette you familiar with this game let's see some sort of you know mmm not practically I mean you know you probably wouldn't be as successful as you are in your careers if you were engaging in this game often so but basically the Russian Roulette game is a game of chance allegedly popular with officers of the Bolshevik Army stationed in places like Siberia where my understanding is that there was not a lot of things to do so in order to get some excitement that would take revolver that looks a little bit like this they would place one bullet into the one of the chambers so the spinet take pets instead of you know put it together press the trigger and I will generate sort of you know not journaling boost excitement I would win some money as if they actually want so they wouldn't be bored anymore or the gang would go off and again they wouldn't be bored anymore so win-win situation so obviously we can't play this game because of health and safety reasons and also sort of you know shooting each other is probably not good business practice so let's play it a third version of this game so in my third version of this game to make it a little bit more extreme I'm gonna ask you to think about a revolver with six chambers and I'm gonna place three bullets in the inter chambers in consecutive chambers okay so six chambers I'm gonna place three bullets in consecutive chambers and I'm gonna give you a choice this is your data science assignment would you rather sort of go for a spin once and then shoot shoot shoot or would you rather go for B which is spin shoot spin shoot spin shoot so in both cases you are shooting three times okay and in both cases there are three bullets in consecutive chambers a question is what your rather to go with a spin shoot shoot or B spin should spin should spin so now I think about it for a few seconds well I have some water it's very important to sort of structure this break who things who thinks a spin shoot shoot shoot is that either choice 1 2 3 4 5 good things spin should spin should spin shoot is that I choice 1 2 3 4 5 6 I would call it around 25 plus or minus 5 because that kind of the level of confidence I have on my ability to count numbers that are greater than 5 when I teach the designs are typically start by playing this game with the with the audience and typically the split is something like you know 20 25 percent of the people go for optional a and sort of 75 to 80 percent of the people go for option B so who went for option B like wants to sort of explain quickly why they did this yes sir okay so you know the first time I spend us like a 50% chance of getting into sort of a loaded chamber which is not so good and a 50% chance of surviving so the first time I do this I have something like you know half or 3 out of 6 probability of surviving times 3 out of 6 times 3 out of 6 because every time I spin provided I survived the previous time I set the probabilities to sort of a half which gives me a probability of surviving which is 1 over 8 okay that's that's I agree with this analysis what about optioning yes sir context experienced yes [Laughter] all right so let's let's analyze this a little bit better I don't know if it's the Russian part of you are so they're you know background or the mathematics part not help but there's certainly sort of some truth to this so let's think about sort of you know the chambers of the gun so if I spin the gun and I land on chamber number one what happens bank two back three Valentin six Clickbank five click click but if a land of four let's click click click so there is one chamber out of six that if I happen to be lucky enough to London this I survive for sure and then another five where I die either immediately or after a little bit of suspense so basically are the chances of surviving here is 1/6 and so the video clip are sitting in terms of percentage that's something like a 12.5% and this is something like what is it sixteen point six six seven something like that so in a sense relatively speaking this one has a 25% chance higher of surviving or if you want to think about this in terms of investment where return on investment will be a lot of times if I percent higher if you chose a rather than B okay so why am I giving you this little assignment you think well what am I trying to get out of I don't take to communicate I guess the first point I'm trying to communicate is that our intuition which has been sort of you know developed over to the you know millennia of life in a Savannah where we had to show them you know hunt for animals protect ourselves and our offsprings from attacks from lions so that we refined partners to sort of you know pass down our genes and so forth the evolutionary forces that developed this intuitive understanding we have about this world are not particularly good in making trade-offs that involve probabilities so our intuition yes is great for many many things but not necessarily for or sort of you know making choices about investments that involve risk uncertainty especially dynamic ones that have all their time so so so that's kind of the first point the second point is that you know a little bit of analysis gets you a long way so what did we do for analysis here well we built a little model of a gun and if you look at this model of encanta does this look like a gun let's cut the shape of a gun the weight of a gun does it fire bullets no it simplifies a lot of the things that make the gun what it is but retains Daewon feature of a gun that's relevant for the problem at hand and this is the chambers of the gun and the ones that are loaded and the ones that are empty now based on the simplified version of reality which we typically call a model we did some analysis okay very simple analysis once we had this model and based on this analysis we reach the conclusion and in this case we also communicated this conclusion to what I would say was a pretty hostile audience to start with an audience I kind of disagreed with so the what I thought was the right course of action so so in a sense for me this is the essence of data science data science always has to start with sort of an end goal a decision something you're trying to achieve it built sort of choose a or choose B it builds a simplified version of reality that allows us to fine tune our intuition to some analysis to calculation and communicate this calculation to achieve buy-in and consensus to action so if you think that this is sort of relevant and this is something worth doing then I think you will like learning more about data science and hunting your understanding of it um so in other words data science is this attempt to tame the animal spirit and for this I have this quote by one of my heroes John Maynard Keynes who was an economist based at Cambridge so he said that most business decision are taken as a result of animal spirits of sport dan you search towards action rather than inaction and remember I mean option number B has a lot more action to it right you're spinning shooting spinning shooting spinning spinning as opposed to action a which is like you know in action you just spin once and you sit passively hoping for the best instead Maynard Keynes believes that decisions should be made as the outcome of weighted average of quantitative benefits multiplied by quantitative probabilities so this is data science timing the animal spirits okay now you may be thinking that this is all nice and good if you're playing little sort of you know games of chance for the experiments and so forth but how much value that is really out in the real world well let me try to sort of address this head-on by telling you a story and the story involves a US retailer called target so I haven't worked with target I haven't worked for target but I have worked for one of their biggest equivalent I would say lore among competitors Tesco's in the in the UK so I've done a lot of work for them so the story for target starts with sort of the the father of a teenage daughter working into their log out Tesco store and asking to speak to the manager so the father was clearly very upset so the manager comes to speak to him and and the father expresses how dissatisfied he is with target because they send his daughter promotional material for pregnancy and he was upset because he was accusing target of glamourizing teenage pregnancy and they are for corrupting his daughter's morals and demanded to know why target thought that it was appropriate to send the material like this to a 17 year old girl I mean the manager apologized profusely agreed with the customer that this was inappropriate and promised to investigate to find out what has happened and in doing the investigation here's what he found out he found out that instead of you know bug at the central offices the data science team had created a pregnancy prediction tool so the tool was targeting people's purchasing activities and it also linked to public data about a baby delivery so in the US whenever a baby's delivered this becomes public information and it created a pregnancy index so for example when somebody starts buying unscented lotions certain food supplements when they stop buying alcohol when they used to in the past and so forth these are all signs that that person is more likely to be pregnant now why would target that want to do something like this well I mean there is that when somebody becomes pregnant especially for the first time they start buying a lot of stuff they were not buying before like anything that has to do with so the me no pregnancy fashion to preparing for their arrival of the baby to sort of after the baby arrives purchasing a sort of you know anything from diapers to sort of you know bottled milk and so forth and why does it important to know that somebody would be a pregnant for a target well because if you know well in advance you can start sending promotions you can start so the you know capturing part of this wallet that's going to be opening wide for the arrival of the new baby I know I have an 18 month daughter so you know he can be expensive and be ahead of the competition because the moment the baby is born this is public information and that inundated with offers so once the algorithm so that would predict somebody had a high probability of being pregnant then target would start sending from ocean all materials for example discounts for a pregnancy wear diapers and so forth now how do you think that customers responded when this happened at the at first they click on it okay okay actually that wasn't their first response a little bit uncomfortable with this I would say that's even a euphemism they freaked out I mean they were like you know my parents don't know I'm pregnant how does target know I'm pregnant Anderson Amy from also my period so actually at the beginning it rather backfired they're insisting so they tried a different approach and the different approach was to sort of you know send Barry the promotions for pregnancy stuff with other promotions that will clearly be relevant to a pregnant woman like wine glasses or a loan moaner so that made it sound like the promotions were random and therefore it was less so the you know spooky history let's big brother-ish if you like and much more successful in converting demand from so promotions to practices so so basically this is exactly what has happened at Target and out of this predictive modeling they managed to sort of gain a large part of the sort of baby Maternity market and it was one of their fastest sku categories over the last decade now back to my story sort of you know week later the manager called the Father to apologize again for Dane appropriate promotions that his daughter was sent but this time the father was not so angry he was actually a little bit apologetic so he said that he had a conversation with his daughter he found out about some events that have been happening in his house called that he was not aware of so this was a case of target knowing before the father that he thought I was expecting okay so how did a target manage to unleash value out of this predictive analytics and I think you know in general what does it take for a company to create value I think it's helpful to think about it in three pillars you need to have the right infrastructure this is the infrastructure that allows you to capture store retrieve and analyze data so in target's case they had a very detailed customer purchasing database and for every customer they managed to create a unique customer ID that linked to the credit cards they were using their loyalty cards and they were also able to link this to public records of births and deliveries using this data they were able to sort of create a more complete understanding of who is likely to be pregnant and when in fact they were so good at it that they were able to predict the delivery day will give or take twenty days and if you look at the best medical models of predicting delivery based on sort of G stations and scans and so forth there are only slightly more accurate I know because I Healthcare is another AG welcome so okay so the first things it's to have the infrastructure the second is to have the tools in the methodology available this is the tools that allow you to describe and understand the past that is which of your past customers were pregnant and how they are purchasing habits changed predict the future that is you know from the customers that you have now that you don't know if they are pregnant or not which of them are pregnant and when are they going to deliver I also provide insights and prescribe actions so what kind of promotions are they more likely to respond to and they're going to be more profitable to us and even better if you can automate this called decision so that once a certain behavior is observed in the data without any human intervention a certain action certain promotion is triggered these are all things that were able to do well but I would say perhaps the most important pillar is the third one which is the culture and the processes so target marketing department went to the data science team with this challenge can you come up with an algorithm to predict pregnancy so if the marketing department if they have the fourth sight the intuition to understand that such a model would be viable this would have never happened then sort of you know once the data science team came up with an algorithm that was predictive pregnancy's they had the right systems in place to deploy this solution and start sending out promotions and perhaps the most important thing they didn't just start up you know stop there once the first solution was deployed because the first solution was kind of failure they iterated they continued improving both the predictive models but also the way they were using these models to generate pike and I think in most organizations if data science fails to make an impact is usually the last part I mean questions are not asked solutions if they are asked and their solve solutions tend to be deployed in a transactional manner we're sort of you know the first solution is rolled out and then we forget about it and move on to solving something else without any feedback loops that are out for continuous improvement let me contrast the a target story with the story of another organization I work with at the moment and this is Anna nature's hospital so the Vera great stuff organization start by well-trained people that do their best to provide service to sort of you know the critical services I should say to the population so one of the very first projects I did with them was sort of trying to help them decide the bed assignment that is the hospital a go scenario on a relatively rare to the video a situation where it had some extra money and therefore it wanted to expand some of its units so it asked the units to a bid or make proposals for why they should get the money to have more bets what do you think happened well you know the surgery director I'd put in a request for more surgical beds and there our justification was that surgeries were the most profitable operation the hospital was doing so by having more surgeries that would increase the hospital's revenue more than they would increase the costs and therefore it was the right sort of investment to be made the emergency department they wanted more birds because emergency diamond was growing by 7% per year for the last three years and they just were coping with the amount of demand they were having maternity services they wanted more beds because target campus has been so successful in glamorizing teenage pregnancy so anyway well I'm trying to say it was pretty much a screaming match and who got the beds the emergency department got the beds why because the government was targeting waiting times for emergency departments and if they didn't get the best the hospital would be embarrassed it would be bad publicity and they would be kind of penalized so what was missing from this I mean nobody asked to see the data nobody asked to see the evidence why well the infrastructure was kind of there but so they were collecting information but not with the mindset of using it to inform operational decisions more with the mindset of using it for compliance and reimbursement so for example the the data they were collecting didn't have much information about costing for example and if you want to use their data for sort of you know any sort of personal improvement costing is kinda important to some methods I would say the hospital had about 200 people with the with the title data analyst what was the job of these people to ensure that the data entered the databases correctly and it was used correctly for reimbursement so they were getting paid for the work they were doing and maybe a little bit of descriptive analytics so try to describe what has happened in the past how much growth the hospital had seen and so forth they had very little capability of forecasting future demands and even costing different services that the hospital was offering to figure out where where the biggest means but but most importantly the culture was missing I mean the hospital was run I would say by experienced managers that have been doing this for a while but everyone had their own idiosyncratic style of managing and there was a lot of repeat and trial and error so in a sense the hospital was missing the culture of asking data questions the odd ology to answer them and even the infrastructure to answer them was kind of incomplete now I'm optimistic about hospital sir I think this is the area where data science will make a huge amount of impact in the near future but I just want to contrast a sort of you know the retail setting with where this is much more advanced to sort of you know the healthcare and the hospital setting where all of these are kinda missing okay so you know you may be thinking okay I'm not in retail I'm noting hospitals so what does this mean for me well I would say that data science has a role to play and an impact to make in most areas of business from sort of understanding your customers better - so in order to sort of you know attract retain and better serve their needs - pricing managing promotions figuring out what products to offer where especially we are working on a multi-channel - the retail setting a huge amount of value from data science can be gained from organizing internal operations more efficiently from supply chain management inventory planning even operations like website design and website optimization all of this can be informed through data science HR management is one of the frontiers where data science is making a big impact right now some people think not necessarily for the best for example with artificial intelligence doing pre-screening of CVS in order to identify right candidates leading to all sorts of issues around discrimination and perpetuating biases that have been prevalent the historical data aren t a new product development so the amount of data science put forward in R&D for let's say pharmaceutical products is staggering a clinical trial optimization and so forth is a very active area of data science but even companies like Zara in designing and deploying a new products they have been using data science very extensively and in fact one of my colleagues in the operations department of LPS is working with them extensively to improve on this so what I'm trying to say is that pretty much the limitations of data science are the limitations of our imagination as sort of data scientists and managers let me also sort of continuing this with a sort of you know taking data science a level further so in all of these cases in all of these sort of examples have been thinking about it and talking about data science is used to do what the company has been doing but a little bit better a little bit more efficiently a little bit more proactively to generate more value now the next step to this is organizations that take the ability to the data science well and people the whole business model around this so the biggest example I have for this is so the you know the case avocado I mean I suspect you know what the kado is I'm a pure online play retailer so we'll go back into retailing which by virtue of being purely online it had a very good understanding of its customers because they did all of their purchasing on in digital channel so their purchasing history goes very much sort of recorded because it didn't have the burden of legacy physical stores it had a blank slate on which it could design are the fulfillment centers and Depot's this is a picture of a double where you have robots working on a grid to try to order to try to to actually assemble customer's orders and take them to the point where they can be loaded in a van and be sent a customer to sort of you know optimising truck loading and truck rooting through busy cities with traffic and so forth so at some point became so good understanding and predicting customers needs and designing websites that were really fit for purpose and helping customers through the retail journey that became really good and really efficient at online fulfillment and reducing the cost of fulfillment so much with their optimization tools okay let's become a technology company instead of selling groceries let's sell technology that enables other organizations that don't have such a big footprint in online to create this really quickly and I think the response of the stock market to this is kind of telling so they our market capitalization yesterday was eight point four billion pounds on sales of 1.6 billion and actually no profit they've never made a profit contrast this with a much more established player like Tesco whose sales are 50 times or 45 times larger by this much a capitalization is only you know three times larger so in a sense by pivoting to become a technology company the this is this this loss making retailer has managed to attract valuations associated with sort of you know artificial intelligence startups now you may think that this is sort of reflecting true value of data science so data science has generated all of this volume people think that beta that business model to harness data science is the source of this or it could be a bubble and in fact there's a lot of evidence to suggest that their recent bubble in artificial intelligence investment so for example last year there was estimated nearly four and a half thousand our funding deals involving startups with the actual size of the amount of funding going to startups running in the hundreds of billion dollars a lot of it by the way happening in China and you know I wouldn't necessarily disagree with this sort of notion that this could be a bubble that is we have exhibited expectations of what data science and artificial intelligence can do leading to another flow of money into the industry leading to valuations that are not and think with fundamentals in other words a bubble butt pad is really a problem and in sort of you know thinking about whether this is a problem I think I'm saying two questions becomes important the first question is where there are these exuberant expectations whether they're financing technological innovation versus financing speculation and I think in this case it's clear that the investments are financing new technology rather than speculation the second question is whether this financing is done via equity or debt and again in this case most of this funding is equity rather than that so contrast this with the subprime mortgage bubble there we had a problem of foreign enhancing speculation using debt so when the bubble bursts there was a big spillover in the real economy and the actual investment amounted to sort of a little improvement while when you are investing in something that creates a technology like the dot-com bubble did in essence that left a legacy of sort of the internet companies like Amazon or Google and search engines okay that the backbone of sort of the smartphone technology that were using now and so forth and when it did burst because a lot of it was I could define ants they didn't have a huge negative impact on the rest of the economy so it to the extent that it is a bubble I think it's a good bubble yes [Music] so it is a problem to the point where I creates too much hogwash yes too much noise and people would start saying why should I go into something people who are not aware of what it is and how to use it I think it's a concern but I think any backlash will be transient and their value the fundamental value to the extent that you believe it exists will be validated like exactly the dot-com example you mentioned is a very good luck so there goes is sort of you know polish for a little bit of time but you know fast forward 15 years later 20 years later the most valuable companies in the world are you know the apples the Google's the Facebook's and so forth whose business model is data science so if this bubble bursts and a lot of these startups lose money is their ability to generate value by design is going to be diminished no I think that's gonna be diminished it's like suberin expectations so it would be recalibrated somewhere that's closer to reality may be very calculation for a little bit is below reality it will very quickly catch up so I'm not I'm not so much worried about this I think actually the legacy of this bubble bursting when it bursts will be new methods a bunch of very well trained to the people that can take jobs in more productive industries and perhaps sort of you know a wider acceptance of data science and artificial intelligence as a driving force of the economy this is the split across different years I mean just by eyeballing their way back the relative colors here it's not identical abuse it's actually growing so it's a less of a red here more more maybe little less here so it's not the same but and I'm bear in mind that these are estimates so up by the way most of my slides are if you go in the corner I put references which is where I got this information form so I mean you can you can look into this most of my slides because I look at me right it's not my research I wouldn't vote for the methodology either but I think it's indicative of where they were [Music] in a sense in any way some in data scientists that are much more comfortable talking about methods so he wants to talk about networks and how you do propagation but I find the business model of artificial intelligence and how it's perceived in popular media and culture fascinating in ways that you know nobody knew what I was doing you know ten years ago when I was doing it and now it's kind of every I think wants to talk about it so it's investing it so I find this fascinating so I do want to sort of you know switch to sort of a little bit of sort of you know that they start trying to downplay this a little bit as well because you know I did build up the case that data science can help sort of you know with multiple areas of business and can even transform business models but it also has limitations and I think it can to acknowledge this so for example I mean if you take one of the sort of you know biggest current topics in the UK brexit how good is the science and AI predicting brexit predicting the impact of it happening I and if you think it's stable it's terrible the data science and AI is not good at predicting these things well how good are we how good are you this is there some questions out there that are fundamentally not well suited for data science to be answered to be it to be answering and if I want to sort of you know give a sort of framework to thinking about what types of decisions could we think about throwing at data science I think it's very helpful to think about it in sort of these two dimensions the first has to do with the complexity of the decision we are trying to inform so in a sense in the target case the situation was to try to predict whether somebody's pregnant or not it's a kind of binary decision so therefore complexity level is relatively low compare this with say you know driving a car driving a car involves multi-ball highly complex decisions so do I want to accelerate do I want to show the you know turn three degrees to the left then accelerate then take ten degrees to the right and then the accelerate if I wanted the accelerate by how much how do I make what would I need to do now in order to avoid the pedestrian that I think was going to cross the road in front of me and so forth so so the complexity of the decision space the second dimension I think is very helpful to think about is how strong feedback loops are and think about it like this so if I make a mistake you know in predicting somebody is pregnant do I find out yes I mean the public records would say if somebody I predicted is pregnant you know gave birth after you know nine months and if you know ten months have passed nothing else happened I probably know I made a mistake you know contrast is and similarly with driving a car I mean if I sort of you know drive a car and I cause an accident do I know that the custom accident yes can I criticise the sequence of events that led to its accident happening and sort of you know change the code so the next time I find myself in similar situation I will avoid this yes so contrast that with say the situation of recruitment okay so if I don't hire somebody do I find out if I make the right decision not to hire them no because even if I hire them it takes quite a bit of time to know if that person develops to sort of the the leadership position I was hoping they would I mean sometimes it very really bad fit I find out very quickly but I mean the difference between a good and an excellent sort of you know Canada it takes such a long time to find out that I would say the feedback loops are weak so if you're in a situation where you have load decision complexity and strong feedback loops this is the area where data science is a no-brainer you should be able to make an impact and you should be able to automate decisions so that a lot of the things that we're done in precisely slowly manually through trial and error could be automated and become a sort of precise and a lot a lot of life if you're in a situation with high decision complexity but strong feedback loops like self-driving cars I think there's hope and the hope is that there's a technology of data science and artificial intelligence improves so that you know more and more things that we think are too complex now we'll start becoming sort of you know sufficiently low complexity to be able to make an impact but you know it's a first sort of you know recourse I would say the decision augmentation should be possible through AI and artificial intelligence so think of all of the driving aids that are becoming available you know you know of our cards now a standard so that this is a first step towards trying to automate the process completely so the you know on the other extreme where we have weak feedback loops so therefore we cannot train algorithms to make good decisions and there's always going to be a subjective human element to the decision making I think data science has a role to play and this is to support decisions by providing sort of recommendations perhaps highlighting salient features that a human operator may have missed perhaps crunching a lot of complex information to reduce it to something that more it's more digestible but it's unlikely that an automated system will take over the decision-making and if you are in a world of high decision complexity with weak feedback loops like I don't know politics or writing novels or designing welfare systems I think it's not even helpful to think about data science so just forget it okay so I would always start from sort of you know this corner go down up in order to understand where to look for bugs all right so we talked about a lot of things let me just also try to bring it up to what does it mean for me and by me I don't mean me as a data scientist by me I mean somebody who is a manager who is sort of you know data literate but perhaps not a data scientist so I would say buying large data science for you is an opportunity it's an opportunity and not a threat why because I don't think it will replace human intuition and experience even in the areas where they're low complexity with strong feedback loops you still need a strong amount of human intuition experience and contextual understanding in order to make sure you are asking the right questions you're implementing the right approach to solving them and you've built the feedback loops that allow for continuous improvement but I think it's important to learn how to think like a data scientists so so you could hire an expert but you need to be able to talk to them so therefore I think it's becoming increasingly important to understand what data science methods can do for you and I'm thinking of things like data visualization supervised learning for our predictive and classification models and supervised learning to sort of you know try to see patterns in data that are not obvious things are run simulation and optimized to allow to squeeze out the efficiency of systems even estimating a models are based on neural networks and deep learning which are so the video models inspired by the way the human mind is a wired the era of data science being something that's done in a basement in an organization it's kind of over I think they designs is sort of you know rising up through the ranks of organizations very quickly and in order to unleash its value for companies but also for personal career development I think it's important to invest in these skills that's why I'll be as we're developing a large number of electives for our MBA students because we think is impossible for MBA student to consider their education complete without at least a high-level understanding of all of this so let me finish with some homework like I finish all of my lectures my colleagues and I maintain a website on the London Business School review that talks about analytics and big data so we talk about trends we see and how big data data science artificial intelligence is trying to make an impact in the world both the good impact and the perhaps not so good impact so so by all means go have a look we will make these slides available so if you want to look at any of the references that I've been talking about you can just click on the slides but I want you to sort of you know finish this session by reflecting whether there is one opportunity for your organization where data science can create a value so what is this opportunity what are the barriers that prevent your organization for creating this value why haven't you captured by out of data science in this particular opportunity yet what are you missing is it the culture is it the processes is this access to methods is it the data infrastructure and what you do to resolve this and it would be great if Adele PS we could do something to help you unlock value out of data science either through the courses we offer but also perhaps more importantly through the research we do because the research we do tries to sort of generate new methods that solve problems that have nothing solved before or apply existing methods to new problems in innovative new ways that in addition to solving the problem generate knowledge so I would love to hear from you and what the problems your organization's facing are and what we can do to solve them so please stay in touch among the most social media I also actually check email as well so so so thank you and it's been an absolute pleasure speaking with you for the last hour or so thank you [Applause]

Channel: London Business School

Views: 2,147

Rating: 4.8000002 out of 5

Keywords:

Id: V3mEFk8cD9Y

Length: 48min 35sec (2915 seconds)

Published: Fri Jun 26 2020