Adjusting Attributes

In this section I’ll explore combinations of attributes for dealign with overplotting that occurs with a large dataset, or in particular, when regions of high-density

Inappropriate adjustments — shape & alpha

Inappropriate adjustments — position & alpha

Appropriate adjustments — size & alpha

Small points, with alpha

log-log model

At this point it seems appropriate to return to a univariate distribution to explore this trend more closely. In the figure below we can see the spikes in count at specific carat values but also the sparseness in values at the low end. That is , the negative space in out plot also contains information and we only see that when the bin width is small enough.

Small-width histogram 1

Here, we can also examine differ sub groups

Small-width histogram 2

Group by size bin widths and calculate price: