It’s absolute numbers, which are meaningless without context.
If there are ten times as many Pitbulls as Rottweilers, this chart would indicate that Rottweilers are twice as fatal.
If violent psychopaths are disproportionately likely to own Pitbulls and train them to sic people, there will be more attacks from that breed. If psychos prefer Pitbulls because of their tenacity and jaw strength, more attacks will be fatal.
Breeds that don’t clamp on might be more high-strung and likely to attack, but not cause fatalities. What if Chows are responsible for ten times the attacks that pits are, but they result in survivable maulings? Would that be okay because at least they didn’t kill anyone?
Lastly, fewer than 19 deaths per year in a country of over 300 million is very close to insignificant. Around fifty people are killed by lightning each year, another fifty by stinging insects, another fifty by hot tap water.
The point is, your scaremongering graph contains no useful information. (And the carved dogs really are amazing.)