Interp1D

1D interpolation methods.

Dependencies

No dependencies

Interpolater

Base interpolater class. Not to be directly instantiated. All interpolaters have the same interface; linear interpolation is shown below as a demonstration:

import interp1d

x = [1, 5, 17]  # Inputs
y = [20, 6, 14]  # Outputs
mapping = interp1d.Linear(x, y)
mapping(11)  # 10.0

The input x array must be sorted.

By default, querying a value outside the range of input datapoints will raise a ValueError.

mapping(18)  # raises ValueError

We can control this behavior with the fill_value argument by either specifying the string "clip" or by specifying a tuple of 2 floats representing (low, high):

import interp1d

x = [1, 5, 17]
y = [20, 6, 14]

mapping = interp1d.Linear(x, y, fill_value="clip")
mapping(0)  # 20
mapping(18)  # 14

mapping = interp1d.Linear(x, y, fill_value=(-100, 100))
mapping(0)  # -100
mapping(18)  # 100

Linear

Linear interpolation.

Cubic

Cubic interpolation.

MonoSpline

Monotone preserving cubic interpolation. Unlike Cubic, MonoSpline guarantees interpolated values to be in-between it's supporting points.

The above graphic comes from antdvid's implementation, which this implementation is also based on.

searchsorted

Not an interpolation function, but may be externally useful. For a given sorted iterable, finds the index where a value should be inserted to maintain order.

data = [1, 5, 10]
searchsorted(data, 0)  # 0
searchsorted(data, 1)  # 0
searchsorted(data, 2)  # 1
searchsorted(data, 8)  # 2
searchsorted(data, 11)  # 3