Mr. Hemanth Anumandla began his guest lecture by giving an overview of data analytics. He said he was in the business of telling compelling stories from data. He said that there are reasons for the poor adoption of data stories. Some of them include lack of actionable insight, information overload, ineffective use of visualization, longer time to insight, poor UI design etc. So, how do you write compelling data stories? First, make the emotional, social and cultural connections between the data and the user. Second, include information about the who, what, when, where and how of data and then consider the user’s point of view.
The process of creating the data stories has the following steps:
- Setting the objectives and defining the problem and what to visualize
- Understanding the user and customizing the problem to the user’s domain
- Identifying the constituent entities, listing the information data and insights.
- Categorizing the entities and choosing relevant encodings.
- Choosing a narrative theme
- Using design principles and adopting guidelines
- Engaging the audience
- Iterating, improving and improvising
Mr. Hemanth then analysed the ROI of a media spend by a marketing company. He then moved on to understanding the user. Taking the example of a product, he said some of the key questions which are to be discussed are – who are the users of the product and why they use it, what actions can I enable for them, when and how frequently they use the product, any current challenges that need to be addressed and are there any users with special needs.
Mr Hemanth said that there should be a clear-cut strategy for understanding the user clearly. Some of the patterns of questions that the company looks answers for are how they are performing, how they are tending, how do the trends compare and if there are any problems to be resolved.
He then went into technical aspects of data such as dimensions and metrics. According to him, dimensions are descriptive attributes. Some of the dimensions that describe the data are geographic locations, divisions, products, groups etc. Metrics are individual elements of a dimension that can be measured as a sum or ratio such as revenue, costs etc. Identifying the dimensions and metrics is necessary for understanding data. After this, the data analyst should map user objectives to metrics. If we take an example of a marketing company, for a marketing head, the objective is to evaluate the effectiveness of media spend. The metrics for this is finding the overall spending and sales, distribution of spending etc.
He then moved on to the concept of visual encoding. The various data points which can be conveyed through visual encoding are the proportions, hierarchy, flow, patterns, range, comparison, relationships, distribution, data over time, text analysis, geographic information etc. These data points are usually ordered by their perceptual hierarchy. He then gave us some guidelines on visual encoding such as listing what we want to convey in the order of importance, shortlisting the pre-attentive attributes than can be used for the above relationships, validating the attributes and mapping them to the messages.
Since the objective of the data analyst is to find the stories behind the data, Mr. Hemanth concluded what has been an enthralling lecture by saying “Think Stories not Charts”.