Cashflower

Added caching functionality to Model Point Set's `get()` function to improve runtime.

Refactoring of code connected to creating a graph away from `Model` class in `cashflow.py` to `create_graph()` function in `start.py`.

Since this version, it is possible to have model point sets that don't have all the same ids as the `main` model point set.

For example:

main = ModelPointSet(data=pd.DataFrame({
"id": [1, 2, 3],
"premium": [100, 200, 150],
}))

fund = ModelPointSet(data=pd.DataFrame({
"id": [1, 3],
"value": [10_000, 15_000],
}))


The `fund` model point set does not have `id=2`.

The model will not throw any error. Instead, it will return `None` if the user tries to read the value.


fund.get("value")


It will return `10_000` for the first model point, `None` for the second model point and `15_000` for the third model point.

This version has changed and simplified the relative imports.
In the previous version, the imports were according to the PyCharm's inspections but failed to execute in other environments.

To adjust your model to version 0.4.4, remove the model's name from imports, e.g.:

- `from mymodel.input import main` --> `from input import main`
- `from mymodel.settings import settings` --> `from settings import settings`

Two new functionalities:

- variables can be defined without parameter `t` (time-independent variables)
- variables can be repeated for all model points (to decrease runtime)

Constant variable:


variable()
def elapsed_months():
issue_year = main.get("issue_year")
issue_month = main.get("issue_month")
valuation_year = runplan.get("valuation_year")
valuation_month = runplan.get("valuation_month")
return (valuation_year - issue_year) * 12 + (valuation_month - issue_month)


Variable repeated for all model points:


variable(repeat=True)
def cal_year(t):
if t == 0:
return runplan.get("valuation_year")
if cal_month(t-1) == 12:
return cal_year(t-1) + 1
else:
return cal_year(t-1)

Changes:

- improved setting calculation direction to the variables