This might be best explained through a scenario.
On April 1st, 10 journeys are completed. They have taken on average 40 hrs to complete. On April 2nd, 5 journeys are completed. They take on average 30 hrs to complete. Repeat the above for April 3rd, 4th, 5th, 6th and 7th, getting the average time taken for journeys to complete on that day. Imagine the following hours were recorded for those: 25, 50, 40, 20, 30 hrs.
On April 7th, the dashboard is now able to start calculating rolling averages, because it has 7 days of data. It will take April 1, 2, 3, 4, 5, 6 and 7 hrs and get the average (40+30+25+50+40+20+30)/7 = 33.57 hrs. So, on April 7th, the point on the graph will show 33.57 hrs (or 1 day 9 hrs to make it more understandable for a business user)
The next day, on April 8th, 5 journeys finish taking an average of 10 hours. On April 8th we take the average times from, April 2nd, to, April 8th, (30+25+50+40+20+30+10)/7 = 29.28 hrs, or 1d 5 hrs. So, on the average journey time graph, it will show 1 d 5 hrs.
The graph builds each day with new data. This is the rolling 7 day average so uses 7 days to calculate the point, but 30 day average will use 30 points.
The reason data may not be displayed at times would be because:
It has not been 7 days yet, or 30 days yet so it can't calculate a true number. It will eventually appear as journeys are completed day by day
All of the journeys are automated and only take a few seconds to completed where it could be the case for some early tenants used just for testing so the number is too small