We are currently lacking a load agent that is "realistically" stochastic. Our current Agents consist of

1. ProfileLoadAgent: Which represents standard load profiles. Which is only really realistic when looking at a multitude of loads in that respective category, since it basically depicts their average behavior.
2. FixedLoadAgent: Which has a fixed load.
3. RandomLoadAgent: Which is kind of what we want, but not really. The problem here is, that the household picks a random value from a distribution. That means the previous value has nothing to do with the last value (we probably should double check, that that is in fact true).

We need a household that is stochastic, but without the limitation of the RandomLoadAgent, meaning that not only the individual values depict a expected distribution, but also the sequence of values is sensible. One modeling approach could be to go for a Markov chain type of process, similar as our half-finished implementation here.