migrate from BSplineKit to DataInterpolations, provide `PolynomialSpl…

…ine` (#178)

* migrate from BSplineKit to DataInterpolations, provide `PolynomialSpline`

* fix compat

* DataInterpolations 4.6 and above requires Julia 1.10

.github/workflows/ci.yml

@@ -9,7 +9,7 @@ jobs:
     strategy:
       matrix:
         version:
-          - '1.9'
+          - '1.10'
           - '1'
         os:
           - ubuntu-latest

Project.toml

@@ -1,12 +1,12 @@
 name = "FinanceModels"
 uuid = "77f2ae65-bdde-421f-ae9d-22f1af19dd76"
 authors = ["Alec Loudenback <[email protected]> and contributors"]
-version = "4.7.0"
+version = "4.8.0"
 
 [deps]
 AccessibleOptimization = "d88a00a0-4a21-4fe4-a515-e2123c37b885"
 Accessors = "7d9f7c33-5ae7-4f3b-8dc6-eff91059b697"
-BSplineKit = "093aae92-e908-43d7-9660-e50ee39d5a0a"
+DataInterpolations = "82cc6244-b520-54b8-b5a6-8a565e85f1d0"
 Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
 FinanceCore = "b9b1ffdd-6612-4b69-8227-7663be06e089"
 IntervalSets = "8197267c-284f-5f27-9208-e0e47529a953"
@@ -29,11 +29,11 @@ FinanceModelsMakieCoreExt = "MakieCore"
 [compat]
 AccessibleOptimization = "^0.1.1"
 Accessors = "^0.1"
-BSplineKit = "^0.16, 0.17"
+DataInterpolations = "^4.7.1"
 Dates = "^1.6"
 FinanceCore = "^2.1"
 IntervalSets = "^0.7"
-LinearAlgebra = "^1.6"
+LinearAlgebra = "1"
 MakieCore = "0.7"
 Optimization = "^3.15"
 OptimizationMetaheuristics = "^0.1.2, 0.2"
@@ -43,4 +43,4 @@ SpecialFunctions = "2"
 StaticArrays = "^1.6"
 Transducers = "^0.4"
 UnicodePlots = "^3.6"
-julia = "1.9"
+julia = "1.10"

src/Contract.jl

@@ -319,7 +319,7 @@ coupon_times(b::AbstractBond) = coupon_times(b.maturity, b.frequency.frequency)
 
 for op = (:ZCBPrice, :ZCBYield, :ParYield, :ParSwapYield, :CMTYield, :ForwardYield)
     eval(quote
-        $op(x::Vector; kwargs...) = $op.(x, eachindex(x); kwargs...)
+        $op(x::Vector; kwargs...) = $op.(x, float.(eachindex(x)); kwargs...)
     end)
 end
 

src/FinanceModels.jl

@@ -11,7 +11,7 @@ using AccessibleOptimization
 using Accessors
 using LinearAlgebra
 using Transducers
-import BSplineKit
+import DataInterpolations
 import UnicodePlots
 using Transducers: @next, complete, __foldl__, asfoldable
 import SpecialFunctions

src/fit.jl

@@ -281,7 +281,7 @@ function fit(mod0, quotes, method::F=Fit.Loss(x -> x^2);
 
 end
 
-function fit(mod0::Spline.BSpline, quotes, method::Fit.Bootstrap)
+function fit(mod0::T, quotes, method::Fit.Bootstrap) where {T<:Spline.SplineCurve}
     discount_vector = [0.0]
     times = [maturity(quotes[1])]
 

src/model/Spline.jl

@@ -3,23 +3,28 @@ Spline is a module which offers various degree splines used for fitting or boots
 
 Available methods:
 
-- `Spline.BSpline(n)` where n is the nth order. A spline function of order n is a piecewise polynomial function of degree n − 1. This means that, e.g., cubic polynomial is a fourth degree B-Spline.
+- `Spline.BSpline(n)` where n is the nth order. A nth-order B-Spline is analagous to an (n-1)th order polynomial spline. That is, a 3rd/4th order BSpline is very similar to a quadratic/cubic spline respectively. BSplines are global in that a change in one point affects the entire spline (though the spline still passes through the other given points still).
+- `Spline.PolynomialSpline(n)` where n is the nth order.
 
 This object is not a fitted spline itself, rather it is a placeholder object which will be a spline representing the data only after using within [`fit`](@ref FinanceModels.fit-Union{Tuple{F}, Tuple{Any, Any}, Tuple{Any, Any, F}} where F<:FinanceModels.Fit.Loss).
 
 and convienience methods which create a `Spline.BSpline` object of the appropriate order.
 
-- `Spline.Linear()`
-- `Spline.Quadratic()`
-- `Spline.Cubic()`
+- `Spline.Linear()` equals `BSpline(2)`
+- `Spline.Quadratic()` equals `BSpline(3)`
+- `Spline.Cubic()` equals `BSpline(4)`
 """
 module Spline
 import ..FinanceCore
-import ..BSplineKit
 import ..AbstractModel
 
+abstract type SplineCurve end
 
-struct BSpline
+struct PolynomialSpline <: SplineCurve
+    order::Int
+end
+
+struct BSpline <: SplineCurve
     order::Int
 end
 

src/model/Yield.jl

@@ -2,7 +2,7 @@ module Yield
 import ..AbstractModel
 import ..FinanceCore
 import ..Spline as Sp
-import ..BSplineKit
+import ..DataInterpolations
 import UnicodePlots
 import ..Bond: coupon_times
 
@@ -33,8 +33,8 @@ Constant() = Constant(0.0)
 FinanceCore.discount(c::Constant, t) = FinanceCore.discount(c.rate, t)
 
 # used as the object which gets optmized before finally returning a completed spline
-struct IntermediateYieldCurve{U,V} <: AbstractYieldModel
-    b::Sp.BSpline
+struct IntermediateYieldCurve{T<:Sp.SplineCurve,U,V} <: AbstractYieldModel
+    b::T
     xs::Vector{U}
     ys::Vector{V} # here, ys are the discount factors
 end
@@ -60,12 +60,28 @@ function FinanceCore.discount(c::Spline, time)
 end
 
 function Spline(b::Sp.BSpline, xs, ys)
-    order = min(length(xs), b.order) # in case the length of xs is less than the spline order
-    int = BSplineKit.interpolate(xs, ys, BSplineKit.BSplineOrder(order))
-    return Spline(BSplineKit.extrapolate(int, BSplineKit.Smooth()))
+    order = min(length(xs) - 1, b.order) # in case the length of xs is less than the spline order
+    xs = float.(xs)
+    knot_type = if length(xs) < 3
+        :Uniform
+    else
+        :Average
+    end
+
+    return Spline(DataInterpolations.BSplineInterpolation(ys, xs, order, :Uniform, knot_type; extrapolate=true))
+end
+
+function Spline(b::Sp.PolynomialSpline, xs, ys)
+    order = min(length(xs) - 1, b.order) # in case the length of xs is less than the spline order
+    if order == 1
+        return Spline(DataInterpolations.LinearInterpolation(ys, xs; extrapolate=true))
+    elseif order == 2
+        return Spline(DataInterpolations.QuadraticSpline(ys, xs; extrapolate=true))
+    else
+        return Spline(DataInterpolations.CubicSpline(ys, xs; extrapolate=true))
+    end
 end
 
-
 include("Yield/SmithWilson.jl")
 include("Yield/NelsonSiegelSvensson.jl")
 

test/Yield.jl

@@ -111,8 +111,19 @@ end
         maturity = [0.5, 1.0, 1.5, 2.0]
         zero = [5.0, 5.8, 6.4, 6.8] ./ 100
         zs = ZCBYield.(zero, maturity)
-        @testset "$i" for (i, curve) in enumerate([fit(Spline.Cubic(), zs, Fit.Bootstrap()), fit(Spline.Linear(), zs, Fit.Bootstrap()), fit(Spline.Quadratic(), zs, Fit.Bootstrap()), fit(Spline.BSpline(5), zs, Fit.Bootstrap())])
-
+        variants = [
+            Spline.Cubic(),
+            Spline.Linear(),
+            Spline.Quadratic(),
+            Spline.BSpline(5),
+            Spline.BSpline(3),
+            Spline.BSpline(1),
+            Spline.PolynomialSpline(1),
+            Spline.PolynomialSpline(2),
+            Spline.PolynomialSpline(3),
+        ]
+        @testset "Constructor $i" for (i, m) in enumerate(variants)
+            curve = fit(m, zs, Fit.Bootstrap())
             @test discount(curve, 1) ≈ 1 / 1.058
             @test discount(curve, 1.5) ≈ 1 / 1.064^1.5
             @test discount(curve, 2) ≈ 1 / 1.068^2