
Using matrix scatter plots for a quick overview
What happens if we want to visualize a lot of scatter plots in a single graph to quickly get a sense for the data? In that case, we need matrix scatter plots. We have various package options to create such matrix scatter plots (such as the car package). However, to keep things simple, we will use a built-in function instead of an external package.
By looking at the graph shown below, we can get a big-picture view of the interactions among variables. The purpose of this type of visualization is not to provide details, but to provide a general overview. To read this plot we need to look at any interesting scatter plot in the matrix, and move both horizontally and vertically until we find the name associated with its axis.
For example, if you look at the plot immediately to the right of NoQuals and simultaneously immediately on top of L4Quals_plus, what you're looking at is at the relation between those two variables (NoQuals in the y axis, L4Quals_plus in the x axis), and we find that it's an inverse relation; the higher the percentage of people in a ward with high levels of education, the lower the percentage of people with low levels of education. Another obvious relation is that the higher the education level (L4Quals_plus), the higher the occupation (HigherOccup).
Matrix scatter plot
Due to space restrictions, we were not able to show all variable relations, since the scatter plots would be too small to make sense of. However, we encourage the reader to add more variables to the matrix. There are some non-obvious relations. Finding them is left as an exercise for the reader:
desired_variables <- c( "AdultMeanAge", "White", "Owned", "NoQuals", "L4Quals_plus", "Unemp", "HigherOccup", "Deprived", "Proportion" ) pairs(data[, desired_variables])