Data Visualization: When Does it Work and When it Doesn't?

Data Visualization: When Does it Work and When it Doesn't?

Data Visualization (DV) can sometimes be very useful as it may show something that's hidden when we only see the raw data. Many inferences can be drawn by visualizing data and that can be really useful in many conditions. But on the other hand sometimes data is just visualized without any proper reason and that is not adding any information to the present situation.

How can you assess if data visualization will work or if is it not needed?

INTERPRETABLE
First of all, to get valuable insights from the data, it should be interpretable and relevant. So the data you are using should be collected and structured as most of the data available is random and unstructured, this will help you decide if the data is interpretable.
RELEVANT
Secondly, the data should be relevant to the problem you are facing.
USEFUL PERSPECTIVE
Finally, it should be original and should focus on the problem at hand and give a new perspective.

If the information doesn't fit the above criteria, then DV cannot make it more useful. So probably the efforts won't be worth to visualize the data.

Reasons to visualize data (Advantages)

CONFIRMATION: If we already have a set of assumptions or intuitions about how the current situation is or how the business is going in some area, then visualization can help us confirm those assumptions. We might compare the current situation with previous year. Or visualizing the data can help us make better decisions if the assumptions prove to be wrong.
EXPLORATION: When we have a mountain of data we can use visualization to segment that data from various perspectives and to build and improve a model that will allow us to predict and improve the system. This practice of using visualized data instead of statistics is known as Exploratory Data Analysis (EDA). It can be used by the companies to make decisions with more confidence.
EDUCATION: There are 2 main ways in which DV can be educational:
- To simply the reporting (e.g., the comparative charts and graphs part of the DV from the raw data).
- To develop intuition and insights based on the changes and evolvement in the data. Here insights are developed by visualizing the data over time and observing the changes.

Risks in Data Visualization (Pitfalls)
There are 3 main risks in using DV. Our ability to control them defines the value of the visualization we are creating. Ignoring them can reduce the impact of DV.

DATA QUALITY: The data available to us may be unstructured and sometimes we may not know the source and reliability of the data. But data quality is very important because it is the raw material for the whole process of DV. So if bad quality data goes in, it will result in (an amplified) bad quality of DV which in turn may lead to bad decisions or the visualization won't make any sense. So it is important to check whether the data is reliable and complete.
CONTEXT: The whole point of DV is that we could point out some new and useful insights from that. So we have to check all the potential relationships between the data elements. Leaving some of the contextual data untouched may diminish or destroy the insights that can be drawn from the data.
BIASES: The person who will visualizes the data may influence the visualization through using particular color combinations, positioning of data elements, and sometimes using visual tricks, any of which can challenge the interpretation of the data. This can significantly influence the understanding of the those who use the visualization.

Source: Jim Stikeleather, "HBR Guide to Data analytics basics for managers", 2018, pp. 171-182