 My name is Balakrishna Chamola, I work with TCS as a senior user experience designer. Me and my team provide data visualization design solution to all our stakeholders within TCS. And we take care of all the data visualization activity that is starting from right from understanding of its need, kill the conceptualization part of it. How do you do it? Yes, know it in a form of ingredients and recipe. People come to me with huge list of data and ask me to create visualization that should best represent their data. Some come to me with predefined list of reports and visualization and ask to create a very creative visualization out of it which is already been available, but it needs to be meaningful. So, over a period of time this became routine of mine and one fine day I was sitting and I was thinking about it. So, I felt as if I am a chef or I am a cook and they comes with basket full of ingredients and ask me to cook some delicious or wonderful food out of it. So, that is how I started relating food with data visualization, ingredients with the preparation part of it and the recipe with the method of it. That is how I derived to a conclusion or I arrived to my title like design ingredients and recipe for data visualization. A colleague of mine came to me one day, he was a non-UX guy and he got bored of his monotone stuff. He came to me, he said, hey, Bala, you guys do a great stuff and creative stuff, which I could be a designer, a data visualization designer like you people. I said, why not? You can be a designer. When I looked at him, he was, there was a big question mark on his face. He said, how? How is that possible? How can I be a designer? Because I do not have any background of it. I said, not a problem. Like to create a good food, you need to know its proper ingredients and recipe. Similarly, to create a good data visualization, you know to its relevant ingredients and recipe. He said, wow, is there something available, then please share with me. I wanted to be a designer. I said, fine, let me take you to our kitchen and I showed him the menu that we have for him. So, first you need to understand what is data visualization, why data visualization is needed and why it is important. Second, the ingredients, third the recipe and then I have shown him some of our dishes. So I gave him a brief understanding about data visualization, like things that you get from a textual format data or a tableau format data. You cannot, that, things that you cannot derive it from them. You can derive it from a data visualization. Like from a data visualization, you can uncover and you can identify more thing. You can derive more insight from data visualization, but that is not possible when you see data in a tableau format or in a text format. What data visualization gives you? It's a pattern, trend and correlation of data. When you see pattern, it's talk about the various behavior of data, like where the data have been cluttered together, where you can find some outliers of those data. So those are kind of pattern you can derive from a visualization. Trend, you see a behavior of data over a period of time, how they are performing. Third is the correlation, how the data are related to each other, how they are connected. If they have some relationship, they are connected or if they are disconnected, you will find those data away from it. So these are the key things. There are many things that we can derive from data visualization, but these are the main things that you get it from a data visualization and not from a normal format of data or tableau format of data. So to make this concept clear, I told him about, I gave him an understanding about a traditional chart, basic chart, and about trend visualization. So even basic chart represent data visualization, right? They represent data, how they are performing and all that. But things that trend visualization shows you, it cannot be shown through a traditional chart like, again, in trend visualization, you can interact with data, you can explore the data. Whatever a different individual want to see from the data, they can play with the data and they can explore it and they can drill down to more level so that whatever they inside they want to infer, they can get according to their need. But same thing you cannot do with the traditional chart and graph. Is it fine? Yeah, this is really cool. I wanted to know, now I'm curious to know about the ingredients and the recipe part. Said, let's go ahead. So I took him to the ingredient section. So in ingredient section, what do you need to know? When you get a requirement of data visualization, it's not necessary that you have to, whatever data you get, you start, directly start building up the visualization, it's not like that. You need to analyze the data first, understand whether there is a need of a visualization. If there is a need of a visualization, then you define an objective of this visualization. When you say objective, what comes under objective? In objective, you need to define what is the question that the visualization is going to answer? Who your user would be? Or what are the tasks they are going to perform? Whether they are going to monitor data, they want to see a performance of something or they want to track some information. So those come under the objective. So once you have defined your objective, you evaluate and identify the parameters. So he asked, what is parameter in context of data visualization? I said parameters over here means a list of data points, which club together or set of data points when they are clubbed together and it derives one inside or one part of visualization to it. It's a parameter over here. So if you need to define a parameters if they are not available, but if they are already parameter available, evaluate those parameters. See whether those parameters are meeting the, meeting user's objective or not, or meeting the visualization objectives or not. So once you have list of those parameter, bucket them and map parameters. And so bucketing the parameters which are correlated to each other. You can group them together and put into one single bucket. And if you found multiple bucket and you see that still there are parameter which can have some common connection amongst them. You can map those parameters in a different bucket. In this exercise, you might also see there are some parameters which are not relevant or which are adding noise to your visualization. They are redundant, so get rid of those parameters. But if you feel that at some point of time they might be useful, group it in a separate bucket. So again, it's eliminating cluttering parameters. Then you move on to create narrative part where you create a story out of this parameter that you have identified. It will help you to identify the problem statement that you, based on which you will create a visualization. And then it will also give you an overview to the detail view of the visualization, how it needs to be designed. So he was fine with it. He said, okay, this is fine, I understood this. Now, why don't, so he was more excited to know about the recipe part. So there are three steps in the recipe, like understanding your data measurement scale, knowing the right type of charts and graph. And third is the understanding the graphic variables. So when we say data visualization, the output of data visualization is a graphical representation of data. To a high level, there are many great definitions to it, but just for a quick understanding, it's a graphical representation of data, but it needs to be meaningful. So we have data and it needs to be represented in a visualization manner. So definitely, there should be some visual encoder, which will be related to the data. That's how we'll be representing those data in a visualization manner. And these graphic visual encoders have some properties, they have a property of qualitative and quantitative. So some graphic variables represent qualitative kind of information, and some other graphic variable represents qualitative kind of representation. So to know that, which type of data needs to come under qualitative and quantitative, you need to know its data, what are those behavior of those data, to which category they fit into. So that's why we start from a first step, knowing your data, to which category does they fall. So step one. So there are four type of data measurements scale, nominal, ordinal, interval, and ratio. So it has been also termed as not, that's why you can see an OI and initial letters have been kept as bold. So nominal and ordinal, they define the category of a data. This measurement scale have data which cannot be measured or which cannot be quantified. For an example, a male female, these are two different category, you cannot measure it. So it comes under nominal. Nominal are also called as categorical data. Similarly, ordinal, ordinal talks about the sequence or order of a data of a category, like a rating scale, or a ranking from low, medium to high. If these are the order of a data, that falls under ordinal, whereas interval and ratio, by the name itself, we can figure it out that it's a numerical kind of data. Intervals talk about the time interval of data, whereas ratio is purely about a number, a digit of numbers. So since nominal and ordinal represent the qualitative part of data, a category of data, it comes under qualitative property. Likewise, interval and ratio, since they talk about number, they measure numbers, it's talked about the quantity of a data, it falls under quantitative. So this guy was very happy. Now he want to move on to the next step. He was very curious to know how to design a visualization based on all this concept, but still there were two step to it. So the second step is the selection of chart and graph. So this is a high level view of the purposes of charts and graph. There is an extensive list to it, but I have just kept a summarized version for your quick reference. So there are five purposes. 98% in visualization, you will find these five purposes. It might be relationship, comparison, composition, distribution and trend. And sometime in visualization, you might find combination of all these purposes and at that time you might find only one purposes from a visualization. So for example, for relationship you can show it with the help of a network diagram. For comparison you can show it with a bar graph or column graph for composition pie chart and so on. Then comes the third step which I talked about the visual encoder of data. So when you need to properly decide what type of graphic variable you need to choose for your visualization. So for that you need to understand the basic there. Again for the graphic variable component or visual encoder, there is a huge list of graphic variables but this is again a summarized list which we have used, which I have chosen to be shown for this presentation. So from colors, position, texture are the main graphic variables that are used for this for showing a data visualization. And again for the graphic variable also needs to be mapped with their respective qualitative and quantitative measures. So you see color shape orientations, different colors, different shapes and different orientations. It gives you a view that they are not identical. Each and every element is talking about different things. So that's why it's coming under qualitative and whereas in qualitative quantitative if you see the size of bubbles from smaller to big you can easily relate it that a smaller size means a less amount of data and bigger size means big amount of data. Likewise it goes for other graphic variables in under quantitative. He said, fine, I can see some confident in this guy. Now he said, but I'm not sure how I will be able to cope up in a real life scenario. So come on, let me show you one of our list. So out of multiple use cases, I have taken one use case from bus service planning which is a kind of a smart city kind of project and I have used the simpler visualization to show him so that he can easily understand how we have to apply this ingredients and recipe in a real life project. So you see the insight that was needed from this use case is to measure the number of people, passenger waiting at a bus shelter and the average wait time of a particular bus shelter and the best and best performing bus shelters. So when we extracted that we did ingredient parts what we saw a list of we found a 17 data sets which were related to this visualization but not all the 17 would be showcasing a visualization. So you need to analyze this data and it got classified into three different category. At the bottom you can see these are the shortlisted variables which will represent your actual visualization whereas the top part you can see those are part of filters or some other kind of interaction. So it becomes a part of an interface whereas the insight variables becomes a part of your actual visualization which will represent your data. You have a question? So these were the data that were extracted from the ingredient parts the exercise that we did while analyzing the data and this was the final set of data set we have extracted from the requirement. Yes this data into visualization. So we classified those data not all 17 would be a part of visualization then we classified it. Then as a part of recipe how do we do it? Step one is to map this data with the respective data measurement scale and that's how it's fallen under bucket of qualitative and quantitative so that we can define what type of graphic variable needs to be chosen. Then what we identified that this visualization is needs to be shown trend comparison in relationships that was a part of a requirement. But when we thought of using a traditional charts and graphs we felt that it might add clutter to the visualization or we may not able to show all the insights together. So we thought of going with customized visualization. Then to show customized visualization we need to choose the respective graphic variables with the data that we have categorized in a step one. So here you see the green side, green is the qualitative and blue is the quantitative. We see the data that have been placed under qualitative which will mapped with qualitative and data under qualitative will map with quantitative. That's how you choose how to create a visualization. So this was the output that we came out of it and you can see the green color indicates the good performing bashelter, the red color indicates the worst performing bashelter as the bubble size indicate the size of the people at a bashelter and the arc shape is indicating about which is in a metaphor of time it is indicating the time. Now this guy was confident and I thought of showing him some more dish to build is more confident level very high. So this is one of the visualization that we did where you can see the social network and participatory surveillance app which we have and the sensory devices have contributed towards reducing the manmade disaster over a period of time and another visualization from a passenger boarding which was again related to that project. And third, where you can see a different merchant channel transaction has been happened through different channel like critical, debit card check and their subcategory. This guy was happy and he said, wow, I'm very much thankful to you and now there was a new beginning to his career and he started as a designer. That was just a designer. Thank you. Thank you.