Data Visualization Strategies - petermr/CEVOpen GitHub Wiki

Any data visualization will basically depict one or more data attributes in an easy to understand visual like a scatter plot, histogram, box-plot and so on. This page will cover both univariate (one-dimension) and multivariate (multi-dimensional) data visualization strategies. The code for the various strategies will be covered in the github.

The type of data visualization strategy applied to the results of ami.gui must depend on both the type of attribute as well as the dimension of the data.

Univariate Data Visualization Strategies

Univariate analysis is most simply analysis of data analysis, which is only concerned about visualizing one dimension of data and analyzing the sole attribute.

One way to visualize all numeric data and its distributions, is to make a histogram using pandas. These plots include visual depictions of the nature of the data, for example the skew or the frequency distribution of the data. A histogram is an essential tool when trying to communicate the distribution of the data to others.

When to use a histogram:

1.The data is numerical 2.Viewing the shape of the data distribution, especially when determining whether the output of a process is distributed approximately normally 3.Analyzing whether a process can meet the user’s requirements 4.Analyzing what the output from what ami.gui's process looks like 5.Seeing whether a process change has occurred from one time period to another 6.Determining whether the outputs of two or more processes are different

When visualizing discrete, categorical data, bar plots and pie charts are effective methods to use. However, avoid using pie charts when dealing with more than three distinct categories.

When to use a box plot:

1. When different subgroups of the results of ami.gui need to be compared at a glance 2.To show changes over time 3.The primary attribute is categorical, and the secondary attribute is numerical in nature