Recommended: read the previous article before this one.
Methods to generate dynamic visual for the data containing time as one of the features or when storytelling requires time flow in it
In case of IOT or sensor data stream:
When we are visualizing a dataset with time as one of the feature. It becomes important to plot all the data point with respect to time. For example, if we want to plot a dynamic graph of GDP of different countries or display data of a sensor reading data for a sophisticated IOT set-up.
Scaling the data in the same time frame:
Sometimes in some electronic circuits, different sensors send readings on a particular timestamp with the accuracy of a millisecond. For this problem, we can scale down all the data to the same timescale so we can plot it with one single reference. By this, you can visualize data of many sensors at a one-time frame.
Visualization of the live stream of data
Performance of the system:
The hardware which is being used and its utilization play an important role when it comes to the performance of the system. In this era of cloud computing, this problem is solved but not everyone can afford the cost of the cloud service. In the case of the small-sized businesses, cost efficiency is the key to maximum profit.
A simple example of this situation is stock prices. It is constantly changing with time and think about what if the system is slow and you are getting the delayed results. It can be a disaster or can make you a fortune. In short, it will not help you with your decisions.
Making user-friendly and customizable visual:
The low severity issue, the ability of customization. Users should be able to change the colour, shape and size of the graph. Moreover, if there is an entity which has different categories or having a hierarchy then drilling must be available to the users. Let’s understand the word drilling by an example, assume that you are looking at the graph of bitcoin price and current scale is set to month but you want to see the price of two years ago from now then you need a functionality called drilling to set the precision of the graph. These graphs also must be available in all possible formats like PDF, HTML, JPEG, etc.
The first part of the title “user-friendly” is a bit tricky because you have to understand how a human brain perceives the image which is generated by some data. We all have heard that a picture is better than a thousand words. Similarly, the human brain is good with sight than reading through a bunch of data. The brain naturally analyses the graph faster in comparison to the time taken to generate a visual through a piece of code. All thanks to the visual cortex! This topic needs a whole article to describe it properly.
Design aspects of the visuals:
It is a general problem where small utilization takes place like which colour is right for a particular feature representation. For example, if we want to represent that something is going well then the green colour is good for that. However, you can use any colour for that but which one is best is to be decided by the developer cum designer.
The following image shows the graphical representation of a convolutional neural network for handwritten digit recognition.
A trained neural network is just a matrix with weights and biases for different neurons. It is not possible to read a matrix to understand the network but when we visualize it separating each layer we can see how it works. Design of this chart is good enough to represent the whole network. It will become harder and harder when the number of layers in the network increases. That’s where the developer needs to think how to implement it properly to get satisfactory results.
By generalization, people assume that blue colour is liked by most men and pink by the most women. Well, we know that this is not completely true. Still, we use those colours to represent gender. That’s why we often see in a chart where gender categorization is done, the male and female category are represented by blue and pink respectively.
It was a small one but stay tuned. More on the visual cortex in the next part…
Author – Yugant Hadiyal