def add_test_data(scen: Scenario):
# New sets
t_foo = ["foo{}".format(i) for i in (1, 2, 3)]
t_bar = ["bar{}".format(i) for i in (4, 5, 6)]
t = t_foo + t_bar
y = list(map(str, range(2000, 2051, 10)))
# Add to scenario
scen.init_set("t")
scen.add_set("t", t)
scen.init_set("y")
scen.add_set("y", y)
# Data
ureg = pint.get_application_registry()
x = Quantity(
xr.DataArray(np.random.rand(len(t), len(y)),
coords=[("t", t), ("y", y)]),
units=ureg.kg,
)
# As a pd.DataFrame with units
x_df = x.to_series().rename("value").reset_index()
x_df["unit"] = "kg"
scen.init_par("x", ["t", "y"])
scen.add_par("x", x_df)
return t, t_foo, t_bar, x
def scen_with_big_data(self, test_mp, num_params=10):
from itertools import zip_longest
# test_mp.add_unit('kg')
scen = Scenario(test_mp, 'TestQuantity', 'big data', version='new')
# Dimensions and their lengths (Fibonacci numbers)
N_dims = 6
dims = 'abcdefgh'[:N_dims + 1]
sizes = [1, 5, 21, 21, 89, 377, 1597, 6765][:N_dims + 1]
# commented: "377 / 73984365 elements = 0.00051% full"
# from functools import reduce
# from operator import mul
# size = reduce(mul, sizes)
# print('{} / {} elements = {:.5f}% full'
# .format(max(sizes), size, 100 * max(sizes) / size))
# Names like f_0000 ... f_1596 along each dimension
coords = []
for d, N in zip(dims, sizes):
coords.append([f'{d}_{i:04d}' for i in range(N)])
# Add to Scenario
scen.init_set(d)
scen.add_set(d, coords[-1])
def _make_values():
"""Make a DataFrame containing each label in *coords* ≥ 1 time."""
values = list(zip_longest(*coords, np.random.rand(max(sizes))))
result = pd.DataFrame(values, columns=list(dims) + ['value']) \
.ffill()
result['unit'] = 'kg'
return result
# Fill the Scenario with quantities named q_01 ... q_09
names = []
for i in range(num_params):
name = f'q_{i:02d}'
scen.init_par(name, list(dims))
scen.add_par(name, _make_values())
names.append(name)
yield scen
