Introduction to Business Analytics



this is an introduction to business analytics my hope is that this gets your feet wet that you get excited about Butler anymore so little about me before we get started Cody Baldwin I manage analytics and big data projects at HP so I source sales finance and supply chain data for some of our executive dashboards that we use I have an MBA from Virginia Tech I'm a project management professional as well as a scrum master so although I do have a technical background I spend a lot of my time just managing these types of projects ok so we're going to talk about what is Business Analytics how it's done who's using it and then we'll discuss a few challenges and tips that I picked up along the way that be happy to share with you ok so first what is Business Analytics so business analytics involves using tools and techniques to turn data into meaningful business insights so it's using tools and techniques to turn data into meaningful business insights so that data could be from spreadsheets from traditional databases it could also be from Twitter from Facebook and then we're using tools and techniques the techniques might be statistical models or machine learning something and then the hope at the end of the day is that we get some business insights out of it now if you look at this diagram it looks pretty straightforward right but I don't think this does a good job of communicating the complexity of of analytics I sometimes it sort of feels like this to me so we get some good data coming in we also get a lot of messy data that we have to scrub to clean up to make useable and we also get a lot of garbage data along the way that just isn't really helpful at all and then we kind of have to fight with the data battle with it you know to get something out of it it just sometimes it becomes a spaghetti mess it feels like and then because we have you know garbage data or messy data coming in sometimes on the other end we do get some business sites we also get a lot of noise that comes out you know and sometimes if we're not careful and that no laser that garbage that comes out could use to make decisions it can be harmful right so it's this I think does a better job of communicating some of the challenges that that are faced with and when on analytics projects okay so now let's talk about the types of analytics there's descriptive analytics which deals with what happened right so we're looking at historical information to maybe piece a puzzle together an example might be we're looking at data from the previous quarter our sales data for example and maybe comparing that data to the same quarter on the previous year so we're trying to figure out what happened and then there's predictive analytics so that's trying to figure out what might happen so we might use statistical models to do that and then there's prescriptive analytics so that's act that actually involves telling us what we should do so when to get there we might even use some of the descriptive analytics or the predictive analytics to then get to a prescription right we might recommend decisions based on some optimization models that we put together or some simulations that sort of thing in my opinion I think there's all of these are important they have their place but as we go down this list it gets more value-added I think the best type of analytics are prescriptive analytics in my opinion okay just want to talk for a second about some buzzwords so you've probably heard some things similar to business analytics business intelligence decision science data science I don't want you worry too much about it they all really support the same goal and there's true there may be some slight differences but they all are trying to get to the same end right it involves using tools and techniques to turn data into business insights right all of those help us get to that end State okay so now we want to talk a little bit about the process right how we do analytics what's the method what are the methods that we use so similar to the scientific method we always start with a business problem we're trying to solve some of the problems might be how do we optimize pricing what's the optimal price allow us to boost our revenue or how might we segment our customers to tailor product offers to that there's a higher likelihood that they'll buy something or problem might be solving bottlenecks or failure points in our supply chain so we first want to figure out what problem we're trying to solve the next thing we'll want to do is to plan the work so we do some research on the problem we maybe estimate the effort that's required to do the analytics on it and then we'll build a business case to get support from our leadership or whoever's asking us to do this I know sometimes in the excitement of doing the analytics we can sometimes see if this step but I think it's important not to neglect it it certainly doesn't have to take a long time to do this a lot of this can be done in a couple hours but I think it's still important that you don't make any commitments to do something until you've kind of figured out first if it's even possible and then how much effort and money is going to be required to do that so some of the things that we might want to ask about the data that we're needing is first is it available can we even get to the data and if we can what's the quality of the data if it's garbage and we're not going to be able to clean it up and that's going to be a problem and if it's messy we still got to spend time to scrub it so that's something that we need to be aware of another thing we might not want to ask is it granular enough so if we're looking for sales data at the you know city level for example if we can only get into state level or the country level that could be a problem so those are some of the things that we want to think about before we get started right I think it'll save you a lot of pain later if you gone through this type of research in this level of planning now once we've done that we want to put a business case together this doesn't have to be complicated at all can be can be very simple but what we're trying to do is to communicate to our leadership our asses to do the analytics how much it's going to cost and based on the costs what are the benefits that we're going to get out of it and I like to couch these in terms of good better and best options you certainly can make a recommendation which one might be most appropriate or what's the best use of money but it's always nice to give people a couple different options you know I just think it it adds a little bit of value and then obviously once we've shared this with our leadership we don't want to move forward until we have their strong support right we don't want to make any investments until we get there they're buying their approval okay so now once we've identified the problem planned out how we can solve it and gotten you know did some analysis or estimated what the cost might be and what benefit we might get for those costs then so we've got the approval we can start getting to work so what we want to look at doing is grabbing the data that we need finding a place to store it well want to spend some time cleaning it and scrubbing it analyzing it to maybe visualizing it and then interpreting it right on the end we want to get those insights out of it some of these may happen in parallel that's okay or some may be switched around in some cases but you get the idea again we're grabbing the data we're storing it somewhere cleaning it to get into the format that we need and then doing the analysis and maybe some visualizations and then interpreting what we can happen to get those insights okay so next I wanted to share with you some sample architectures so how does this look technically and this doesn't cover you know every way that analytics is done just some of the more common examples that I've seen there's a million possibilities and permutations of this but this just kind of gives you some ideas of some of the more common approaches that are taken the first architecture example that I see a lot actually is what I call traditional analytics so this is for big structured implementations supports that supports lots of demand but sometimes it can be stiff and it's it's designed that way right so there's some governance in place to make sure that the data stays of good quality and that you know it's secure and so only people that see the data should see the data and that sort of thing so it's designed to be a little bit stiff and so it can be more difficult to respond but we do that because we want we need some some level of control some level of Governments right so in this example we're going to pull data from transactional systems so data that's coming from our supply chain maybe from points of sale systems from finance data that maybe eventually gets reported to Wall Street so we're pulling data from transactional systems using ETL so this means extracting the data transforming it getting it into the format we need and then loading it we're going to load it into what we call our data warehouse that's a reporting repository that's where we're going to store our reporting data and then what we'll do is we will point our tools right our our analytics tools on the front end this might be could be Microsoft Excel in some cases or it could be you know some of the new flash your tools like tableau or QlikView we point it to that data to then generate you know graphs or models or whatever it may be okay another example is what I call big data analytics I'm starting to see this more frequently so this is for big structured excuse me big unstructured implementations it supports lots of demand that could support lots of demand really fast right so in this example we might be pulling data from Twitter from Facebook from the web maybe from message boards or sites or something and we're pulling that data into a Hadoop cluster we don't need to go go into Hadoop here but it's talked a lot about when we discuss Big Data but Hadoop allows you to ingest lots of amount of data really fast lots of unstructured data and it'll also also allows you to do some scrubbing on that huge data set that might be really difficult to do in a more traditional environment and ANA cluster is just kind of a collection of computers that are working together so they're all working together to analyze this unstructured data and then we're pulling that data in some cases at least a subset of that data we've maybe now structured it we've cleaned up we've scrubbed in to where data warehouse as we talked about in the previous example this is like a reporting repository right and again we're pointing some of our our tools on the front end whatever may be tableau click view Microsoft Excel we're pointing that to our data warehouse my guess that we'll see models like this more and more often as the volume of data that comes in continues to grow and as you see more unstructured data coming in another example is what I call the Excel power-user I don't want to dismiss this at home because this actually happens a lot it still can be analytics right even though you may not think of it that way this model is affordable but it may only be for a limited number of users but it does allow you to respond to needs really fast so an example might be maybe your boss needs you to you know pull in some data from several spreadsheets or from Microsoft Access database that they have some and maybe maybe a little bit of data from the web I want you to do some research analysis try to solve a business problem that you can do in Excel Excel allows you to do it it's not a huge data set you know and you can easily when you're done when you've whatever when you've done your and your pivot tables or put your must difficult and Excel or you can even do optimization in Excel to once you've done that all you have to do is then email out the file share your your findings and be done with it I see this a lot and it's still certainly can be analytics okay now I want to look at a case study that I really like from Capital One now they're not the only one that does this but they're one of the first people to try it out so what they do what they did was take publicly available credit and demographic data so this is data that you get off the web right or where it wherever right and they figure out build models to determine how to segment customers and then what products to tailor them so based on someone's demographics where they live and how much income they make or their credit history based on historical data they can figure out what those types of people bought and then tailor future offers to new customers so in this case what happens is if you know as a customer calls in or as they call a customer right before that customer even says allo that models running in the background against that customers information and they're figuring out okay based on what we saw in the past and based upon who this customer is we think they are most likely to buy this type of product and we ought to offer them this type of credit limit and what's interesting in this model is that it was able to predict what a caller might want to buy with greater than seventy percent accuracy that's huge so you're not just taking this really broad approach right your tailoring offers very specific offers to certain types of customers a really cool example there's million others this is just one example that I that I really like okay now we want to talk about some of the challenges that you might face the first is the data often needs a lot of cleaning we've talked about this a little bit already but what I found is sometimes it's not so much the analysis this difficult it's actually getting the data ready cleaning it and scrubbing it and getting it in a format where you can do that analysis this can sometimes be the most frustrating part of it right just having to clean the data to get it ready another challenge is that data is getting much more unstructured so it doesn't all come from relational databases or from spreadsheets we could be getting PDFs music files pictures movies we even talked about data that's coming in from Twitter from Facebook it's getting much more challenging right it's in a variety of different formats and to add to that data volumes are growing much faster so not only is the data a lot of it messy it's coming in in a variety of formats and it's coming really fast traditional models you know like ones that we mentioned are getting having difficulty handling that I like this quote from IBM 90% of the data in the world today has been created in the last two years it's just going to continue to grow at pretty alarming rates and so on top of all this we need to keep in mind as well is that markets are becoming more competitive right it's demanding decisions faster if we want to stay competitive so not only do we have more difficult data and more data to work with we've also got to be able to you know produce insights at a faster rate in order to compete another challenge good analytics doesn't always solve bit bad business process so if we have a process that that's producing poor data that we're using for our analytics the insights that come out of that might be meaningless or even harmful if we use them to make important decisions so if we have garbage coming in we might have garbage coming out only silver lining to this is that based on the the garbage that comes out you might then realize that you've got to make some adjustments to how that profit you're collecting prop and data on the process that's the only you know silver lining that that this you might want to think about okay just as far as a couple tips go the first one that I think you ought to keep in mind is that you always ought to be providing recommendations and not just data or information right so we get insights that come out and we can use those to help provide recommendations to our business anyone can just summarize and graph data on excel there generates and pivot tables maybe but real value comes when you can provide actionable recommendations to your business so obviously supported by that data that you've that you've analyzed another tip and this is becoming more important as I see this come up more frequently is you want to allow your business more freedom to explore the data that you bring in a lot of our business partners are becoming very tech savvy and they want to do some analysis on their own right it used to be that a lot of this work was done in silos so you had database administrators and maybe statisticians and the business wasn't as actively involved but they want to be involved right they want to be able to do the analysis where we can come in is by providing them some guidance and some governance just so it's not the wild wild west right and the retrieval tools that they can use to make discoveries on the room another tip is to not dismiss expert advice or your gut so some people think that analytics might eliminate the need for expertise right or experience but in a lot of cases they can support each other or supplement each other or confirm each other so it they can work together really well it doesn't have to be one or the other another tip you should break problems down into smaller pieces and deliver data insights more often so you don't have to have solve the entire problem before you start sharing some of this information you can build build excitement as you start sharing some of the insights that come out you know and if sometimes if we wait too long to share information it might be too late to make an optimal decision another's hip keep a data dictionary this is just something to describe your data talks about the database tables that you have that of a schemas or the sources that you're pulling from the fields and the tables and the database keys the data types the frequency with which you get the data the granularity how it's being loaded and this will help other people as you know come after you right who are looking to solve in this problems and and don't realize that maybe the data is already available and so they may just need to look at your data dictionary to figure out if it fulfills their requirements another tip don't get too rigid with your architecture so we talked about how data is becoming more complex it's much more unstructured it's coming in a faster rate and we have to have a variety of tools to be able to handle some of this so it's not that you know hammer is just not going to work it's got to be a toolbox right several things that we may need to use and sometimes the tool that we use depends on the type of problem we're trying to solve or how soon we need to get some insights out and one final tip I think it's so important to make sure you master pivot tables and some of the statistical tools in Microsoft Excel I'm not saying that you're going to use that for all analytics but it's a great way to start exploring some of the data that you get to do some of the research and planning that we've talked about earlier it's a really great way to start slicing your data up and to start exploring it ok so that's it I hope that you enjoyed it you got excited about analytics that you emerged in learning more thanks

34 Comments

  1. mayur rindhe said:

    awesome content. please make playlist in depth about Business Analytics

    May 22, 2019
    Reply
  2. hounddog00 said:

    Nice overview! Thank you, Cody.

    May 22, 2019
    Reply
  3. Ayse Girisgen said:

    Found so much value in this video, thanks Cody!

    May 22, 2019
    Reply
  4. Shalahuddin Al Ayyubi said:

    where I can learn more about this topic?

    May 22, 2019
    Reply
  5. Darrell Mayson said:

    Loved your vid. As someone who can easily get carried away and blinkered to whatever 'tool' or process has caught my fancy your gave me a higher level appreciation of the data analytics biosphere. Confirmed a few things from my own experience and provided food for thought. Thanks =D

    May 22, 2019
    Reply
  6. LK LL said:

    it's been 3 years, what happen Cody? do you have new videos?

    May 22, 2019
    Reply
  7. Michael Nylif said:

    Really Cool!

    May 22, 2019
    Reply
  8. ncardez said:

    I have been doing this without knowing some of the things that you mentioned, there's a lot of noise and buzz words around analytics, just now I'm trying to see what's the trend with the "science" word added in data, the only thing I have not been doing yet is presenting statistic for predictions, and I guess that is the tricky part and therefore the science word involved.

    I'm studying again some maths concepts to fill this gap that I already have.

    Good video!.

    May 22, 2019
    Reply
  9. Ïrthu Suresh said:

    Why did he stop making videos? He is really good at this ….

    May 22, 2019
    Reply
  10. Michael Pisciarino said:

    0:54 What is Business Analytics?
    2:38 Types of Analytics: Descriptive, Predictive, Prescriptive
    4:38 Business Problems (Optimal Price, Customers, Bottlenecks/Failure points)
    6:45 Business Case: Cost And Benefit

    May 22, 2019
    Reply
  11. sudhir sale said:

    Such a informative video 🙂 thanks mate.

    May 22, 2019
    Reply
  12. Vidal Trejo said:

    Excellent video, one of the most clear and very well designed videos i ever watched,

    May 22, 2019
    Reply
  13. Stats Work said:

    Hello, You have posted such precious and informative article which gave me lot of information. I hope that you will keep it up and we will have more informative and helping news from you. #Statswork provides best statistical analysis services to the research scholars, students, faculty, corporate, clinical research, and food and beverage industries and so on. For more information Mail to: [email protected] Visit: http://www.statswork.com/

    May 22, 2019
    Reply
  14. Tara Zacharia said:

    Business analytics is one of the most important element to make the more accurate business decision based on data. To become an #Expert #BusinessAnalytics I would like to recommend a complete #Business #Analytics #Course by #XLRI – https://talentedge.in/xlri-jamshedpur/business-analytics-courses/

    May 22, 2019
    Reply
  15. Maranje Odima said:

    Good information !! any books I could read to become certified. I am pursuing an MBA in business analytics and I want to get a professional certification.

    May 22, 2019
    Reply
  16. Balinda Juma said:

    this so incredible like simple explanation

    May 22, 2019
    Reply
  17. Li Zhao said:

    Very helpful. Thank you for sharing.

    May 22, 2019
    Reply
  18. Preeti Singh said:

    How to start a career in business analytics ? Is it necessary to get a certification ? I am a fresher …

    May 22, 2019
    Reply
  19. arun prasanth said:

    I like this vedio very very very much… Thank u…
    I request u to post many vedios regarding business analytics..

    May 22, 2019
    Reply
  20. Abhijeet Gill said:

    Should science students opt for masters in business analytics or this field is beneficial for economics students only??

    May 22, 2019
    Reply
  21. lenin christ said:

    Wow it is really wonderful and awesome thus it is very much useful for me to understand many concepts and helped me a lot. it is really explainable very well and i got more information from your Video.
    Visit a Website
    https://bit.ly/2ILiobL
    https://bit.ly/2ILtxJD

    May 22, 2019
    Reply
  22. Shalini Priya said:

    Nice Video. Thanks for sharing valuable information. it’s really helpful. Who wants to learn this video most helpful. Keep sharing on updated video.
    Visit a Website
    https://bit.ly/2PvabuV
    https://bit.ly/2A7VVmy
    Read my Blog:https://bit.ly/2NFoueN

    May 22, 2019
    Reply
  23. Avirup Sarkar said:

    Can you provide some more case studies? Even if you refer I can google it.. please!

    May 22, 2019
    Reply
  24. Nivatha Sri said:

    Wow it is really wonderful and awesome thus it is very much useful for me to understand many concepts and helped me a lot. it is really explainable very well and i got more information from your Video.
    Visit a Website
    https://bit.ly/2Qb3wWV
    https://bit.ly/2zAdz2l
    Read my Blog:https://bit.ly/2IiWexn

    May 22, 2019
    Reply
  25. Jay Jain said:

    Watch at 1.5x.. good presentation but very slow

    May 22, 2019
    Reply
  26. reenak mahanty said:

    Can a person MSc biotechnology background currently working in pharmrmacovigilance do a analytics course

    May 22, 2019
    Reply
  27. Vasudha Mathur said:

    Crisp… (:

    May 22, 2019
    Reply
  28. Mohsina Kaderi said:

    This is a good introduction to data analysis. But as Excel is an integrated part of analysis data some lecture on excel should be included with it.

    May 22, 2019
    Reply
  29. Fergal Downes said:

    Heya buddy, very nice flick that you have here. Will check back again soon.

    May 22, 2019
    Reply
  30. Rafael Gonzalez said:

    thanks a lot, im currently working with social media analytics and i find this really useful

    May 22, 2019
    Reply
  31. Leon Florence said:

    Obrigado

    May 22, 2019
    Reply
  32. Mauricio Davila said:

    muted

    May 22, 2019
    Reply
  33. tallboywitshortpants said:

    Very educational, thank you!

    May 22, 2019
    Reply
  34. Alex Kimi said:

    Very good presentation. I liked the Tips at the end. They're very useful.

    May 22, 2019
    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *