R Tutorial: Adding more detail to summaries

Key Takeaways

now that you can do simple summaries in this section you will learn some techniques to dive deeper into summary visualizations as a motivating example let's quickly revisit one of the visualizations we just created the distribution of total fair amount shows us that nearly all fares are under $100 but there are a few in the dataset that get up to over $1000 we also see that the average fare is around $10 and some fares are close to zero it would be interesting to dive deeper to see if other variables in our data can explain the variability and fare amount we can often learn more from our data by introducing more variables into our summary computations to see if there are any interesting interactions that arise in this section you will learn a few techniques for doing this when working with two or more continuous variables you can bend pairs of variables in two dimensions and visualize the bin counts to investigate the effect of other variables on summary statistic computations you can group or facet these computations according to other variables and visualize how the summaries vary according to different values of these variables you may have seen examples of grouping as well as faceting in other courses a useful technique for visually investigating the relationship between two continuous variables particularly when the data is large is to count the number of observations that fall into a two dimensional hexagonal grid and plot the result this can be done using ggplot twos DMX function which does the bidding computation for you prior to plotting the granularity of the binning can be controlled by the bins argument to investigate the relationship between the total fare and the tip amount we can make the following plot here we also add a reference line along which the fare matches the tip amount it's surprising that there are many observations along this line where the fare is basically the same value as the tip we see a clear high count linear pattern in blue indicating a standard percentage tip but we also see a lot of variability around that pattern and many fairs with low tips that don't fit the linear pattern faceting allows us to look at data in more detail or in new ways by breaking the data into pieces and applying the same visualization to each subset of the data this is easiest to do when the faceting variable is categorical adding faceting to an existing visualization such as the plot of number of taxi rides versus time is as easy as adding facet wrap to the plot specification and specifying the variable you would like to facet on in this example we facet on the day of the week which allows us to easily observe the evidence strong day a week effect on ridership note that you can facet on multiple variables with facet wrap and another function facet grid can be useful in faceting on two variables note that with faceting any plotting element you add to the plot is applied to all of the facets independently as in this example where we add geometry highlight the differing progression of ridership over the year on different days using a robust linear model ridership is increasing over the course of the year on the weekends while it remains more constant midweek let's now use these