That depends.
I do not know how received data is delivered.
If delivery contains data that is common for all graphs, I would keep data in the document and each view would ask document for needed data to draw view specific graph(s).

If data is sent in chunks separately for each graph, you have two scenarios:
1. You want to safe data in a file for later retrieval. I would still keep data in a document and use document to orchestrate serialization. Store data in a document in some kind of custom data type (structure, class) for views to consume.
2. You do not want to safe data. This poses a very important question: do you need a document object at all? It is for you to decide.