538 now has a model for the outcome of the democratic primaries: https://projects.fivethirtyeight.com/2020-primary-forecast/
They currently predict an average of 1484 pledged delegates for Biden (1990 are needed to snatch the nomination). 1018 are predicted for Sanders, 634 for Warren, 536 for Buttigieg. They give the odds of a candidate winning the majority (1990) of pledged delegates at 40% for Biden, 22% for Sanders, 14% for no one, 12% for Warren and 10% for Buttigieg.
They also have the states. I am surprised on the high variation of the model. To explain this a bit, they model the electorate and then run thousands of simulation with different paramters based on uncertainty in the data. So they have different outcomes and can give a probability. But for instance in 80% of these simulations gets Biden in Iowa between 4% and 49% of the votes, meaning there are 10% chance he gets even lower than 4% and 10% he gets higher than 49%. That is an incredibly broad range and shows a high level of uncertainty. Because of the 15% rule, for the allocation of the delegates there is a bit more certainty.
Remember, this will be updated as new data (polls, endorsements, financials, …) rolls in.
There is an explanation how the model works: https://fivethirtyeight.com/features/how-fivethirtyeight-2020-primary-model-works/