Bar Charts and Pie Charts are used to visualize categorical data. Both types of graphs contain variations as displayed in the visual.
The bars of a bar chart have a couple of key features:
These features make a bar chart super dependable for representing categorical data.
For any chart like a bar chart, the areas we use to represent values must always be equivalent to the relative sizes of the values they represent. Otherwise, readers could be misled and potentially identify patterns that do not actually exist.
Nominal data has labels with no specific order. Thus, we have a lot of creative freedom when choosing where each bar goes on our chart. One way would order our data is by ascending or descending order.
You can do this in Python with the
order parameter in the
.counplot() seaborn method. Within the
order parameter, the
.value_counts() pandas method can order the values in either ascending or descending format. The code snippet below gives an example of how this can be done.
Ordering the bars of nominal data is useful because it makes keys findings, such as the mode of the data, easy to spot. For example, in the graph pictured, we can see that the
resign value is the mode of the
victory_status variable, while
draw is the least common value.
If we are working with ordinal data, we should plot the data according to our categorical variables. For example, let’s say we want to plot the number of students per grade level at a college. We have a table below, which is a preview data from a students.csv file.
We can order the categorical values as
Third Year, and
Fourth Year since they are ordinal. Using
.counplot(), we can input these as a list in the
sns.countplot(df["Grade Level"], order=["First Year", "Second Year", "Third Year", "Fourth Year"])
From the chart shown, we get takeaways that we couldn’t just from looking at the column of data. We can see that the college had an influx of students for the second year class, as
Second Year is the mode (most observed value) of the
Grade Level column of data. All other years have roughly the same number of students. Putting bars in order according to the ordinal data helps us see these patterns and make our visuals informative for readers.
It is crucial that any pie chart you see or create follows these two rules:
Pie Charts have two common pitfalls:
If you ever run into this issue, a bar chart may be the best solution. The picture comparing pie charts and bar charts shows why.
With each pie chart, it is almost impossible to compare separate sectors. However, the bar chart makes the comparisons much easier to decipher.