A lot of convictions are based solely on eyewitness testimony and identification which we know can be very faulty.
As such, such figures are unreliable and suffer from confirmation bias — if your focused myoptically on one version of the event, unless you are very careful in your procedure (which not every police department is), your initial focus can easily lead to a conviction, even if the suspect is actually innocent.
Eyewitness identification of assailants can be wrong in an astonishing 75% percent of cases. This is in rape, where the victim has the most incentive and opportunity (no other people involved, extended period to memorize the assailant's features) to get the identification right, and because DNA is available we can actually test
to see if they are. And they're shockingly poor
— to the tune of 75% wrong. Think of how wrong an identification can get when the eyewitnesses doesn't have as much incentive or opportunity to get the ID right, like the fleeting glimpse of a suspect involved in a shoplifting case.
This leads me to believe that the correlation you cite is very much suspect. It is not a good piece of evidence against the case that the justice system is racist. Even if the majority of crimes are stratified by race, it is the disparity
between the true criminals and the convictions that the racism lies.
A blank incident report for an incident that clearly occurred is not very good PR. An incident report should have been filled with as much documentation as they could whenever a death occurs, even if just to cover their asses.
Something is wrong
with your department if police officers are shooting anyone dead for shoplifting — something is wrong with YOU if you think anyone deserves that fate. And something is wrong with your department if police officers are shooting people even before they knew a crime had been committed.