def guesstimate(
task: list[str],
best: list[int],
expected: list[int],
worst: list[int],
simulations: int,
random_seed: Optional[int] = None,
) -> DataFrame:
"""Estimate the duration of a project based on three-point estimation of breakdown
tasks or milestones using a Monte Carlo simulation.
The project duration is measured in the same unit used to estimate the duration of
the tasks or milestones. It can be a unit of time (e.g. days, weeks) or story
points.
Eeach simulation is the sum of the duration of each tasks picked at random from a
modified-PERT distribution computed using the best-case, expected and worst-case
estimates provided.
In order to make test reproducible, an optional parameter `random_state` has been
introduced.
"""
simulate_project = generate_project_simulator(
task=task,
best=best,
expected=expected,
worst=worst,
random_seed=random_seed,
)
duration = [simulate_project() for _ in range(simulations)]
return compute_stats(duration)
def test_cmd_guesstimate() -> None:
parser = create_parser()
args = parser.parse_args(
["guesstimate", "tests/guess/csvs/tasks.csv", "--simulations", "10"]
)
actual = cmd_guesstimate(args, random_seed=1234)
expected = compute_stats([88, 92, 82, 93, 80, 97, 84, 95, 102, 86])
pandas.testing.assert_frame_equal(expected, actual)
def test_guesstimate() -> None:
actual = guesstimate(
task=["Task A", "Task B", "Task C", "Task D", "Task E", "Task F"],
best=[5, 6, 1, 10, 5, 12],
expected=[10, 12, 13, 13, 7, 25],
worst=[20, 40, 24, 15, 12, 34],
simulations=10,
random_seed=1234,
)
expected = compute_stats([88, 92, 82, 93, 80, 97, 84, 95, 102, 86])
pandas.testing.assert_frame_equal(expected, actual)
def test_guesstimate_best_equal_worst() -> None:
actual = guesstimate(
task=["Task A", "Task B"],
best=[5, 6],
expected=[5, 6],
worst=[5, 6],
simulations=10,
random_seed=1234,
)
expected = compute_stats([11, 11, 11, 11, 11, 11, 11, 11, 11, 11])
pandas.testing.assert_frame_equal(expected, actual)
def test_estimate_normal_no_changes() -> None:
actual = estimate(
scope=100,
velocity=[5.0, 6.0, 10.0],
change=[0, 0, 0],
normal=True,
simulations=10,
random_seed=1234,
)
expected = compute_stats([14, 14, 15, 13, 12, 13, 15, 14, 15, 13])
pandas.testing.assert_frame_equal(expected, actual)
def test_estimate_normal() -> None:
actual = estimate(
scope=100,
velocity=[5.0, 6.0, 10.0],
change=[1.0, 2.0, 3.0],
normal=True,
simulations=10,
random_seed=1234,
)
expected = compute_stats([21, 18, 17, 22, 19, 20, 23, 18, 19, 18])
pandas.testing.assert_frame_equal(expected, actual)
def test_cmd_estimate() -> None:
parser = create_parser()
args = parser.parse_args(
["estimate", "tests/est/csvs/sprints.csv", "100", "--simulations", "10"]
)
actual = cmd_estimate(args, random_seed=1234)
expected = compute_stats([6, 7, 8, 8, 8, 7, 9, 8, 6, 7])
pandas.testing.assert_frame_equal(expected, actual)
args = parser.parse_args(
[
"estimate",
"tests/est/csvs/sprints.csv",
"100",
"--normal",
"--simulations",
"10",
]
)
actual = cmd_estimate(args, random_seed=1234)
expected = compute_stats([7, 7, 7, 8, 8, 7, 8, 7, 7, 7])
pandas.testing.assert_frame_equal(expected, actual)
def estimate(
scope: int,
velocity: list[float],
change: list[float],
normal: bool,
simulations: int,
random_seed: Optional[int] = None,
) -> DataFrame:
"""Estimate the duration of a project based on past sprints velocity and scope
changes using a Monte Carlo simulation.
The duration estimate is measured in number of sprints.
Every simulations is composed by several iterations, each of which represents a
sprint.
By default, velocity and scope change for each iteration are picked at random
following a uniform probability distribution from the provided historical data.
If `normal` is True, the input will be modelled as normal distribution from which
velocity and scope changes will be derived.
In order to make test reproducible, an optional parameter `random_state` has been
introduced.
"""
duration = []
max_sprints = compute_max_sprints(scope=scope,
velocity=velocity,
change=change)
simulate_sprints = generate_sprints_simulator(
velocity=velocity,
change=change,
max_sprints=max_sprints,
normal=normal,
random_seed=random_seed,
)
for i in range(simulations):
rn_velocity, rn_change = simulate_sprints()
duration.append(
compute_duration(
scope=scope,
velocity=rn_velocity,
change=rn_change,
))
return compute_stats(duration)