I am a non-mathematical student, tasked to create a predictive model from a large available dataset, specifically for dynamic and complex (multitasking, chaotic) real-world environments. I will then validate the model with an existing dataset, and perform another study to examine if the predictive model was accurate. As I am delving into the foundation of mathematical models (e.g., Examples of types of mathematical models, https://en.wikipedia.org/wiki/Mathematical_model), I am trying to define the type of model I will have. My question to you is, is my understanding correct?
The IV's that will be modeled include task properties (such as event rate (signal detection), task duration, signal probability), and agent properties (including communication capabilities, gender, age, reliability). The DV's may include any form of task performance (signal detection metrics), subjective workload, and potentially trust measures (in the agent).
Since we are modeling for a dynamic and complex environment, I believe the mathematical model should be dynamic: accounting for time-dependent changes. Is this correct? What should I be taking into account here (e.g., simulation properties)?
Although the most straightforward way to simulation is perhaps discrete modeling, I think continuous would be more appropriate, as the fidelity would be higher (dynamic, complex real-world environment). Any thoughts on the discrete vs. continuous dimension?
Lastly, I thought the model would be deterministic rather than probalistic/stochastic. The dataset will yield fixed inputs, aspiring to project toward a fixed output given the input. Is this correct? What else should I be thinking of regarding this dimension of the types of mathematical models?
Thank you!