4.2 BAR GRAPHS WITH RESPONSE OVERLAY
We can use bar graphs with a response overlay for exploring the relationship between a categorical predictor and the target variable. Figure 4.1 shows a bar graph of previous_outcome with an overlay of the target response. Previous_outcome refers to the result of a previous marketing campaign with this same customer, with most customers not having had such a previous experience.
![Graph of Previous_outcome vs. count displaying horizontal stacked bars for success, nonexistent, and failure (top–bottom). The highest bar is nonexistent. The shades represent yes and no.](https://static.packt-cdn.com/products/9781119526810/graphics/images/c04f001.gif)
Figure 4.1 Bar graph from R of previous_outcome with response overlay.
Clearly, most customers did not have any previous marketing experience with the company (variable value nonexistent). In general, (non‐normalized) bar graphs are useful for showing the distribution of the values of the categorical variable. However, it is not clear which category has the greater proportion of responders. Nonexistent has the most responders but it also has the most nonresponders.
To clarify situations like these, we may obtain a normalized bar graph, which equalizes the length...