Applying Monte Carlo Simulation to a Python Model - Monte Carlo Simulation

Applying Monte Carlo Simulation to a Python Model

Part of the lecture series "Monte Carlo Simulation":
https://nickderobertis.github.io/fin-model-course/lectures/10-monte-carlo-simulation

Full Course Website:
https://nickderobertis.github.io/fin-model-course/


Notes
--------

- It can make sense to set up a separate dataclass for your simulation-specific inputs, or you may add them to the existing dataclass
- Once you start running large numbers of simulations, some unexpected situations may occur in your model such as inputs going negative that were supposed to only be positive, or one input being greater than another when it is supposed to be less. To solve this, we can build functions which produce the random inputs according to the necessary conditions in our model
- Create a function which runs a single simulation, then call that function in a loop over the number of iterations to run all the simulations
- Because we typically have multiple changing inputs and may even have multiple outputs, it is useful to store data as a list of tuples and then create a DataFrame at the end
- It doesn't hurt to take the quantile of the entire DataFrame to see the distributions of the inputs as well. It can be a nice check to make sure your random inputs are working appropriately
- After running a multivariate regression, be sure to add some text interpreting the results
- We can check the standardized coefficients (coef * std) to understand which inputs have the greatest impact on the outputs. Be careful that these results are influenced by your choice of the input distributions. If your input distributions are not reasonable, neither will be the results

Resources
------------

- Dynamic Salary Retirement Model with Monte Carlo: https://nickderobertis.github.io/fin-model-course/_static/Examples/Monte%20Carlo/Python/Dynamic%20Salary%20Retirement%20Model%20Monte%20Carlo.ipynb