Advanced tableau tutorial | Trend Forecasting, Advanced Charts & R Integration | Edureka Rewind

Key Takeaways

the Tableau we are using today is not just for creating traditional graphs and charts you can use it to mine actionable insights thanks to the plethora of features and customization it offers hi all this is upasna from Eda and this session is going to be your stepbystep guide to explore a few lesser talked about options in Tableau for data science and bi beginners we will create calculations to dive deeper into our data to extract insights Trends and forecast through a feature called exponential smoothing we will also create calculations in a few Advanced graphs that go beyond the drag and drop features of Tableau we'll also be looking at how R can be integrated and used in t so without any further Ado let's get straight into the module now the first two modules are more intermediate than Advanced but you'll be needing these to do the advanced stuff I'll be talking about much later in the module so kindly bear with me for a bit so first of all we're going to be talking about forecasting now the forecasting is about predicting the future value of a measure now there are many mathematical models for forecasting and Tableau uses the model known as exponential smoothing let let's move on to our Tableau desktop to see what we can do with it all right now Tableau takes a Time Dimension and a measure field to create a forecast so we'll be doing a simple forecast with which we'll obtain the value of measure sales for next year using this data source we're going to be using this sample Superstore which Tableau provides us with as you can see you have order IDs order date the ship date mode customer name segment country City and so on this is what your entire data set looks like all right so for forecasting first of all we're going to be creating a line chart with our order date in the x-axis and then we are going to be taking sales in the y-axis or the rows and by default this is what Tableau does it is going to suggest the best possible graph if you're going to input any X or Y AIS into it on completing this step you will find this option to set various options for forecast so we going to go into analysis forecast and show forecast it's pretty simple now we going to edit it we're going to keep the forecast length at automatic and fill in values with zero we're going to aggregate by its automatic quarters right here all right you can even choose the timeline from which you want your forecast to be I'm going to choose two years all right the result of the forecast looks something like this you can also get minute details of this forecast by choosing the option describe so you can do this using two methods first is a simple right click on the forecast here you can go to forecast and show describe forecast here you can see the time series measures the forward forecast what the forecast is based on and here is some bonus information for people who are working with analytics here is the initial of quarter 4 2018 right up to the quarter 3 of 2020 seasonal effect Trend and Seasons contribution on your data quality and so on by checking this box you can also see the these numbers in percentages with that let's move on to the next thing the next thing I'm going to be talking about is trend lines this is something important might be an intermediate level operation here but it's going to be useful in the later part of the video so trend lines are used to predict the continuation of a certain trend of a variable it also helps to identify the correlation between two variables by observing the trend in both of them simultaneously here we are back to our Tableau now there are many mathematical models for establishing uh trend lines now Tableau provides four options they are linear logarithmic exponential and polinomial in this module I'm only going to discuss the linear model so again Tableau is going to take the time Dimension and a measure field to create a trend line and we are going to be using the same sample Superstore and we are going to find the trend for the value of the measure sales for next year so again I'm going to drag the auto date here and the measure sales to the row if I go to analysis I see the trend line box is checked in and that is why we can see these lines right here if you touch a line it will show you the sales of the year and the year of order date along with two other values which which are the r squ and the P values if you can see now the r squ and the P values that you see on your screens right now is two measures that is used in the correlation expression just like we did in the forecasting you can [Music] also describe the trend line or Trend model right here this is for you to get the minute details here you have the model formula you have the number of modeled observations filtered observations your sum squared error your mean squared error your R squ standard error and the P value and another thing for people with a statistics heavy background you also have the standard error T value and P value for hypothesis testing the next thing I'm going to be talking about are a few Advanced charts in Tableau now all almost all Tableau users are privy to the various Elementary graphs that you can do by your drag and drop option such charts are easily made using the show me feature of Tableau but since this module is meant for more advanced users we going to move beyond the show me and explore graphs that require some extra computation so first let's take a quick look at what we are going to be making in the next fews sections the first out of which is the waterfall chart all right so waterfall chart basically derives its name from its analogous orientation and flow and we are going to be plotting the running sales of the superstore over its years so let's start with it while I talk you through the implementation so let's begin with a basic graph we're going to take at x axis the order date and at y AIS we are going to be taking profit now because our marks card here is automatic we going to get a line chart a waterfall chart is a derivative of a line chart that is why we are beginning with this graph now I'm not going to be talking in detail about the line chart here for that purpose we have a video on our Channel now this also implies that this is a chart which can be used to analyze the cumulative effect of a measure and see how it increases and decreases as a whole so let's just make it to understand it better so we're going to right click on this profit Bill looking option right here and there's something known as a quick table calculation where we are going to select a running total now we going to change the mark type from automatic to a Gant bar and now we're going to be creating a calculated field so go to analysis create calculated field we're going to be creating something called a negative profit in a bit I'll tell you why are we doing this here we already have profit and it's the negative sign now this is a beautiful thing about tblo if you see in the bottom of the window you can see this message called the calculation is valid if I go back on this and just keep it this way it's going to show that the calculation contains errors now this is a beautiful thing for beginners because you will know where you have gone wrong right from scratch so we're going to go back to what we had written and okay now we're going to drag this column over the size in the marks bar and you will understand why exactly have I created this measure all right now this calculated field was to fill in the space in the Gant chart which was empty a negative value in the profit extended the bar downwards if I would have added a positive profit the bar would have gone upwards now the length of each small bar in the chart represents the amount of change in the profit from 1 month to the next so we're going to put it in month so that we have a bigger chart to work with obviously so finally I'm going to take this profit the actual positive profit and drag it to the color here you can see the darker blocks are the ones that represent more profit while the lighter ones show less profit and by going onto each chart you can see the negative profit and profit of that month and if you want a little variation in this graph you can just go to color edit colors and choose what you like I'd go from red to green red being bad green being good pretty simple pretty self-explanatory and okay now this is your waterfall chart already now the graph that you see on your screens right now could be very easily represented in the form of a bar chart but I'm sure you would all agree that using a waterfall chart was a way more intuitive way of representing the data especially to see the changes in the measures such as a sales profit over the years this implies that this is also a chart which is used to analyze Anze the cumulative effect of a measure and see how it increases and decreases during the whole while with that let's move on to our next chart now this chart is called the petto chart and it is basically a combination of both a bar chart and a line chart in a bit I'm going to talk to you about the significance of it let's go back to Tableau now first of all I'm going to just make a bar chart so that I can talk about a few things in particular so I'm going to be using a subcategory right here in the product I'll take the subcategory put it in the columns and then I'm going to take the sales in the Y AIS I'm going to change the marks into a bar chart all right and then I'm going to sort the subcategories in a descending order please stay with me I have an important point to make here all right now we have it all in descending order so here I have visualized a popular 820 principle of data analytics if you have not heard of it let me try and explain it to you right now so it is often observed that the majority of sales in a Superstore comes from a select few products one cannot expect raw chicken and raw rice to have the same sales figures as cooked Biryani right so this is officially termed as the 80/20 principle meaning that 80% of the sales come only from 20% of the products that are available in the store in our Superstore the principle can be observed in this chart where most of the sales are generated by phones and shares it's a quite popular visualization peretto charts to be honest and hence it is often used for risk management to determine the most common problems that are having the most negative impact on the project but here we will see it can have other applications as well so next I'm going to be dragging over the sales pill into the rows again now we have two charts and I'm going to take the second sales pill and going to add dual access if I turn this into automatic it turns into something like this now I'm going to run a total running calculation right here okay and now I'm going to change one sales pill into a bar graph which is already done and the second sails into a line graph so it basically turned the running total into a line graph as you can see it's a combination of both a bar graph and and a line graph and all that is left to change is the color scheme I'm going to turn this line orange so it creates better visibility and let's just put up some markers on it and it's all possible by using the marks card here right on the left all right with that you have your petto chart ready and next and finally I'm going to be talking about something which is honestly not that difficult to make but it definitely intrigued me enough to make its way into this live which is the motion chart we going back into our Tableau so our aforementioned data set is already imported as we all know and I've mentioned it quite a few times it is the sample Superstore on our xaxis we are going to put up the order date and we're going to change this into month and on our y AIS we're going to put sales and profit both so sales and profit so we have two line graphs and I'm going to be taking the second one and putting it as a dual axis here it appears in two different colors great for visibility all right and if you might have noticed on the right you see this something called measure names right here keep your eyes on the right to see something that's going to happen next so basically the chart I'm going to be making is inspired by Hans rosling's World economic presentation if you all haven't seen it I'd recommend you take a minute after the session and give it a look and by now I'm hoping making trend lines like the one on your screen right now should be easy for you but what we'll be doing is creating this in motion it's kind of like a gif but better so I'm going to be changing the mark type okay I'm going to be changing the mark type into a circle and I'm going to to be dragging the order date into the pag shelf and I'm going to change it just the way it is in the columns which is by month all right now on the right remember I had asked you to keep an eye on the right of the page on the right we have this option called show history which for me is already checked in I'm going to go there go to Trails here the format selected is none but you can go ahead and choose any color or Dash type that you like for me I like solid lines better so I'm just going to keep it that way here I have my speeds selected all I have to do now is play it and you can see it works exactly like a gif and it shows you month by month sales and profit if I want I can increase the speed I'll come really really fast and now how cool is that with that I come to the end of this segment where I had to show you Advanced Tableau charts the next thing I'm going to be talking about is probably the most interesting part of this live which is the r integration with Tableau now R provides a powerful way to do statistical analysis on large sets of data it is also free which is a compelling factor to its growth now because it is an open source new functions and packages are created all the time on it so if you can't find a capability initially you can search for a package that can do it or even create a package of your own now the one thing I like about the newer versions of Tableau is that it's not just a tool meant to create pretty graphs with a M drag and drop option with the release of the the Tableau 8.1 in 2013 came up plethora of new different functionalities the introduction of R to enable making richer and dynamic visualization was one of the most predominant features now R can be used with Tableau for techniques such as clustering prediction and forecasting and these are just a few to name so I wanted to test it myself now I wanted to start the exploration of R and Tableau by doing something really really simple Le like clustering so I used this Ultra popular data set called Iris which I'm sure most of you must have started your journey in r with it contains different features to distinguish between three types of flowers namely virginica Sentosa and versicula so this is what the data set looks like you have your seple length SLE width petal length and petal width in cenm and the class it is divided in into so first let's go through the basics and the installation process before delving into the visualization now for that I'm going to be moving to my R studio all right you basically install something called as an R server for that basically going to do what you do to install all packages install. packages and put these in inverted commas your R serve next you call the library same thing and then finally your R serve then you're going to select it all and run them for me I've already run them once before so I already have these packages installed so I'm going to go back to my Tableau now our scripts are written in Tableau as table calculations which are sent to the r serve package of R and then the functions take place and it is reflected in Tableau so obviously to properly understand and thereby use this feature you must possess at least some knowledge of R and a few different syntaxes now step two you come back to Tableau go to your help option go to settings and performance and manage external service connection here by default it would pick up our serve on server you're going to be typing Local Host and Port it to 631 one one and then test your connection Tableau is going to give you a confirmation like this which is successfully connected to the external server okay all right so now that you have the proper ingredients let's start cooking so basically you make use of tableau's table calculation the one which you find in analysis right here to type your script in R I hope your initial excitement of making clusters is still there so let's proceed my data set has been imported to Tableau and you can basically make this graph while dragging both your petal length and petal width to the columns and the SLE length and width to the rows then I'm going to go to analysis and uncheck this aggregate measure so that I get this cluster like graph I'm going to make this bigger for you all to see so finally to form the Clusters I'm going to take the class which divided uh the flowers into three types and then I'm going to go over to the color what we have here is a scatter plot which shows clusters of data points divided into three distinct clusters now what I'm going to do is I'm going to do the same with r and now compare the two visualizations that we get and for that we are going to be using the most common clustering algorithm I'm expecting to see something like this but we are going to start doing it from the beginning so I'm going to add a new worksheet same thing which I did previously I'm going to take this I'm going to go to analysis and uncheck the aggregate measures all right now I'll be creating a calculated field I'm going to be naming it cluster now these four functions in Tableau are clear to this desktop server that it is going to be used for our script so I'm going to take script for integer and put in the expression so I'm going to be creating a variable called result and I'm going to be allotting a data frame to it we have four arguments and we are going to allot all of it to Cluster now I'm taking off all the columns that we have the petals and the SEL All In centim for all right now all we have left to do is we're going to take the cluster and move it to the mark span it'll take some time to process please be patient all right now let's just change the color a little bit so we can see more clearly going to pick this color and apply it now I'm going to go back to the scatter plot we had created earlier this is the one we had created before and this is the one that we have although there are a few overlaps the two visualizations do appear to be quite accurate and this is only a small gist of the potential of integrating r Tableau its applications are Limitless and I'm sure most of you have already started to think of different ways that you can interact with it now it would be naive of me to say that this is all that there is to Tableau but this is all the demos that I've had for this live session as new versions roll in so do new functionalities not only that people are always experimenting and exploring Tableau and coming up with new visuals there are these multiple blogs where people publish their experiments with data 2 please do check them out also there's something else that I would really want to show so this is it you can also find new and gorgeous visualizations weekly on the tableau's official Gallery page I would definitely advise you to keep up with these posts and create your own visuals and sharing it with the community so I would like to conclude my session by saying see the data show the visual and tell a story stay creative and all the best on your journey as a data Explorer thank you and have a great day