Merge pull request #10 from tbeason/timevarying

add varying coef models

Author: tbeason

.github/workflows/CI.yml

@@ -25,12 +25,12 @@ jobs:
         arch:
           - x64
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
       - uses: julia-actions/setup-julia@v1
         with:
           version: ${{ matrix.version }}
           arch: ${{ matrix.arch }}
-      - uses: actions/cache@v1
+      - uses: actions/cache@v4
         env:
           cache-name: cache-artifacts
         with:

Project.toml

@@ -1,7 +1,7 @@
 name = "NonparametricRegression"
 uuid = "db432338-e110-4b7a-9c53-0ace38eb8f7f"
 authors = ["Tyler Beason <tbeas12@gmail.com>"]
-version = "0.2.1"
+version = "0.2.2"
 
 [deps]
 DocStringExtensions = "ffbed154-4ef7-542d-bbb7-c09d3a79fcae"
@@ -11,7 +11,7 @@ Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 
 [compat]
 DocStringExtensions = "0.8, 0.9"
-StaticArrays = "1.2, 1.3, 1.4, 1.5"
+StaticArrays = "1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9"
 julia = "1.6"
 
 [extras]

README.md

@@ -24,11 +24,48 @@ The two important exported convenience methods are `npregress` and `optimalbandw
 
 ## Examples
 
+### Basic univariate regression
 ```julia
 using NonparametricRegression
 
-npregress
+# Generate some data
+x = rand(100)
+y = sin.(2π * x) + 0.1 * randn(100)
 
+# Estimate with automatic bandwidth selection
+ŷ = npregress(x, y; method=:ll, bandwidthselection=:aicc)
+
+# Or specify bandwidth manually
+ŷ_manual = npregress(x, y, x, 0.1; method=:lc)
+```
+
+### Varying coefficient models
+```julia
+using NonparametricRegression
+
+# Generate data with state-dependent coefficients
+N = 500
+z = rand(N)  # State variable
+X = [ones(N) randn(N, 2)]  # Design matrix: intercept + 2 covariates
+
+# True varying coefficients
+β₁(z) = 1 + z
+β₂(z) = 2 - z^2
+β₃(z) = 0.5 * sin(2π * z)
+
+# Generate response
+y = [X[i, :]' * [β₁(z[i]), β₂(z[i]), β₃(z[i])] for i in 1:N] + 0.1 * randn(N)
+
+# Estimate varying coefficients
+zgrid = 0:0.1:1
+B = npvaryingcoef(X, y, z, zgrid; method=:ll, bandwidthselection=:aicc)
+
+# B is a 3×11 matrix where B[:, i] contains the coefficients at zgrid[i]
+
+# Predict for new data
+X_new = [ones(10) randn(10, 2)]
+z_new = rand(10)
+y_pred = predict_varyingcoef(X_new, z_new, B, zgrid, 0.1; method=:ll)
 ```
 
 ## Detail

src/NonparametricRegression.jl

@@ -16,9 +16,11 @@ export localconstant, localconstantweights
 export locallinear, llalphabeta, locallinearweights
 export optimalbandwidth, leaveoneoutCV, optimizeAICc
 export npregress
+export npvaryingcoef, predict_varyingcoef, varyingcoefficientweights
 
 include("kernels.jl")
 include("univariateopt.jl")
+include("varyingcoefficient.jl")
 
 
 ########################################