With Azure Log Alerts, the 'period' is how far from the query is able to look back in the alert evaluation. For your Alert you have it set to the max which is 2 days, so it would never have uptime that is greater than 3.
You may be able to collect the performance counter for "uptime" and join that to the query for cpu utilization.
The other option is to use a logic app or something to run your query.
Your Friendly Microsoft Employee,
Joe