You have your "Samples to include" set to 10 - so the upper graph is only looking at the 10 most recent samples. You can set it to 0 (which is "ALL") to have the upper graph show all samples in this data collection set, or you can set it to something besides all (like 10) and zoom in on any period. For example, if you leave it set to 10 - and then double-click on the time-graph near the lost packets you're seeing in the time-graph, then the upper graph will focus on that period and you can investigate to see which hops are causing the packet loss (1, 2, or 3).
Also, the escalating latency in your graph between 3:44.50 and 3:45.10 is fairly usual when you're using all your bandwidth (ie: when you're downloading something at the maximum capacity of your connection). That might not have been happening in your case but I see that on my connection pretty consistenly when I'm download a big file.
Hope that helps explain what's happening here. Since I don't know anything about your situation, I can't really comment on anything besides what I see in your graph, and that's rarely *all* of the story.