Rename Model to ScoreDrivenModel (#145)

* Rename `Model` to `ScoreDrivenModel`

* Bump (breaking change)

Author: Raphael Saavedra

Project.toml

@@ -1,7 +1,7 @@
 name = "ScoreDrivenModels"
 uuid = "4a87933e-d659-11e9-0e65-7f40dedd4a3a"
 authors = ["guilhermebodin <[email protected]>, raphaelsaavedra <[email protected]>"]
-version = "0.1.11"
+version = "0.2.0"
 
 [deps]
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"

bench/benchmark_estimate.jl

@@ -7,7 +7,7 @@ scaling = 0.0
 A = [0.2 0; 0 0.2]
 B = [0.2 0; 0 0.2]
 simulation = simulate_GAS_1_1(LogNormal, scaling, ω, A, B, 1)
-gas = Model(1, 1, LogNormal, scaling)
+gas = ScoreDrivenModel(1, 1, LogNormal, scaling)
 gas.ω = ω
 gas.A[1] = A
 gas.B[1] = B
@@ -29,7 +29,7 @@ scaling = 0.5
 A = [0.2 0; 0 0.2]
 B = [0.2 0; 0 0.2]
 simulation = simulate_GAS_1_1(LogNormal, scaling, ω, A, B, 1)
-gas = Model(1, 1, LogNormal, scaling)
+gas = ScoreDrivenModel(1, 1, LogNormal, scaling)
 gas.ω = ω
 gas.A[1] = A
 gas.B[1] = B
@@ -51,7 +51,7 @@ scaling = 1.0
 A = [0.2 0; 0 0.2]
 B = [0.2 0; 0 0.2]
 simulation = simulate_GAS_1_1(LogNormal, scaling, ω, A, B, 1)
-gas = Model(1, 1, LogNormal, scaling)
+gas = ScoreDrivenModel(1, 1, LogNormal, scaling)
 gas.ω = ω
 gas.A[1] = A
 gas.B[1] = B
@@ -66,4 +66,4 @@ gas.B[1] = B
 #   maximum time:     5.374 ms (58.97% GC)
 #   --------------
 #   samples:          3247
-#   evals/sample:     1
+#   evals/sample:     1

docs/src/examples.md

@@ -18,7 +18,7 @@ y = Vector{Float64}(vec(inflow'))
 
 # Specify model: here we use lag 1 for trend characterization and 
 # lag 12 for seasonality characterization
-gas = Model([1, 12], [1, 12], LogNormal, 0.0)
+gas = ScoreDrivenModel([1, 12], [1, 12], LogNormal, 0.0)
 
 # Estimate the model via MLE
 fit!(gas, y)

docs/src/manual.md

@@ -6,7 +6,7 @@ It provides a high-level presentation of the theory behind score-driven models a
 ## Model Specification
 
 ```@docs
-Model
+ScoreDrivenModel
 ```
 
 ## Optimization Algorithms

examples/cpichg.jl

@@ -8,20 +8,20 @@ y = vec(readdlm("../test/data/cpichg.csv"))
 
 # Specify GAS model: a student's t model with location scale transformation
 # (see /src/distributions/non_native_dists.jl in the repository)
-gas = Model(1, 1, TDistLocationScale, 0.0, time_varying_params=[1, 2])
+gas = ScoreDrivenModel(1, 1, TDistLocationScale, 0.0, time_varying_params=[1, 2])
 
 # Fit specified model to historical data
 f = fit!(gas, y)
 
 # Next, we show examples using nondefault keyword arguments
 # Note that we need to re-define `gas` since its parameters were populated by `fit!`
-gas = Model(1, 1, TDistLocationScale, 0.0, time_varying_params=[1, 2])
+gas = ScoreDrivenModel(1, 1, TDistLocationScale, 0.0, time_varying_params=[1, 2])
 
 # In this example, we set verbose to 2, prompting the package to output optimization results
 f = fit!(gas, y; verbose=2)
 
 # Re-define `gas` once again for another example
-gas = Model(1, 1, TDistLocationScale, 0.0, time_varying_params=[1, 2])
+gas = ScoreDrivenModel(1, 1, TDistLocationScale, 0.0, time_varying_params=[1, 2])
 
 # In this example, we set the optimization method to LBFGS with 5 initial points
 f = fit!(gas, y; opt_method=LBFGS(gas, 5))
@@ -39,11 +39,11 @@ forec.parameter_forecast
 # Similarly, we can access the simulated observation scenarios
 forec.observation_scenarios
 
-gas_t = Model(1, 1, TDistLocationScale, 1.0; time_varying_params = [1])
-gas_t_1_2 = Model(1, 2, TDistLocationScale, 1.0; time_varying_params = [1])
+gas_t = ScoreDrivenModel(1, 1, TDistLocationScale, 1.0; time_varying_params = [1])
+gas_t_1_2 = ScoreDrivenModel(1, 2, TDistLocationScale, 1.0; time_varying_params = [1])
 steps_ahead = 50
 first_idx = 150
 b_t = cross_validation(gas_t, y, steps_ahead, first_idx)
 b_t_1_2 = cross_validation(gas_t_1_2, y, steps_ahead, first_idx)
 plot(b_t, "GAS(1, 1) Student t")
-plot!(b_t_1_2, "GAS(1, 2) Student t"; legend = :topleft)
+plot!(b_t_1_2, "GAS(1, 2) Student t"; legend = :topleft)

examples/garch.jl

@@ -30,7 +30,7 @@ ub = [1.0, 1.0, 0.5, 1.0]
 lb = [-1.0, 0.0, 0.0, 0.5]
 
 # Specify GAS model: a normal model with time-varying σ and lag 1
-gas = Model(1, 1, Normal, 1.0, time_varying_params = [2])
+gas = ScoreDrivenModel(1, 1, Normal, 1.0, time_varying_params = [2])
 
 # Set initial point for the optimization
 initial_point = [0.0, 0.5, 0.25, 0.75]

examples/introduction.jl

@@ -1,7 +1,7 @@
 using ScoreDrivenModels
 
-Model(1, 2, LogNormal, 0.5)
+ScoreDrivenModel(1, 2, LogNormal, 0.5)
 
-Model(1, 2, LogNormal, 0.5; time_varying_params = [1])
+ScoreDrivenModel(1, 2, LogNormal, 0.5; time_varying_params = [1])
 
-Model([1, 12], [1, 12], LogNormal, 0.5)
+ScoreDrivenModel([1, 12], [1, 12], LogNormal, 0.5)

examples/nie.jl

@@ -11,7 +11,7 @@ y_test = y[401:460]
 Random.seed!(123)
 
 # Specify GAS model: a lognormal model with time-varying μ, constant σ, and lags 4 and 12
-gas = Model([1,2,3,4], [1,2,3,10,11,12], LogNormal, 1.0; time_varying_params=[1])
+gas = ScoreDrivenModel([1,2,3,4], [1,2,3,10,11,12], LogNormal, 1.0; time_varying_params=[1])
 
 # Obtain initial parameters to start the GAS recursion
 initial_params = dynamic_initial_params(y_train, gas)

src/MLE.jl

@@ -38,7 +38,7 @@ function results(f::Fitted{D, T}) where {D, T}
     estim_results = eval_coefs_stats(f)
     np = length(f.unknowns)
     jarquebera_p_value = pvalue(JarqueBeraTest(f.quantile_residuals))
-    return EstimationStats{D, T}(f.num_obs, f.llk, f.aic, f.bic, np, 
+    return EstimationStats{D, T}(f.num_obs, f.llk, f.aic, f.bic, np,
                                  jarquebera_p_value, estim_results)
 end
 
@@ -47,7 +47,7 @@ function eval_coefs_stats(f::Fitted{D, T}) where {D, T}
     inv_H = inv(f.numerical_hessian)
     vars = diag(inv_H)
     for i in eachindex(vars)
-        if vars[i] <= VARIANCE_ZERO 
+        if vars[i] <= VARIANCE_ZERO
             vars[i] = VARIANCE_ZERO
         end
     end
@@ -70,7 +70,7 @@ mutable struct AuxEstimation{T <: AbstractFloat}
     function AuxEstimation{T}() where T
         return new(
             Vector{Vector{T}}(undef, 0), #psi
-            Vector{Matrix{T}}(undef, 0), 
+            Vector{Matrix{T}}(undef, 0),
             Vector{T}(undef, 0), # loglikelihood
             Vector{Optim.OptimizationResults}(undef, 0) # opt_result
             )
@@ -87,15 +87,15 @@ end
 
 function update_aux_estimation!(aux_est::AuxEstimation{T}, func::Optim.TwiceDifferentiable,
                                 opt_result::Optim.OptimizationResults) where T
-                                
+
     push!(aux_est.numerical_hessian, Optim.hessian!(func, opt_result.minimizer))
     push!(aux_est.psi, opt_result.minimizer)
     push!(aux_est.loglikelihood, -opt_result.minimum)
     push!(aux_est.opt_result, opt_result)
     return
 end
 
-function fit!(gas::Model{D, T}, y::Vector{T};
+function fit!(gas::ScoreDrivenModel{D, T}, y::Vector{T};
              initial_params::Matrix{T} = DEFAULT_INITIAL_PARAM,
              opt_method::AbstractOptimizationMethod = NelderMead(gas, DEFAULT_NUM_SEEDS),
              verbose::Int = DEFAULT_VERBOSE,
@@ -111,23 +111,23 @@ function fit!(gas::Model{D, T}, y::Vector{T};
     # Number of initial_points and number of params to estimate
     n_initial_points = length(opt_method.initial_points)
     n = length(y)
-    
+
     unknowns = find_unknowns(gas)
     n_unknowns = length(unknowns)
-    
+
     # Check if the model has no unknowns
     check_model_estimated(n_unknowns) && return gas
 
     # Create a copy of the model to estimate
     gas_fit = deepcopy(gas)
-    
+
     # optimize for each initial_point
     aux_est = AuxEstimation{T}()
 
     for i = 1:n_initial_points
-        try 
-            func = TwiceDifferentiable(psi_tilde -> log_lik(psi_tilde, y, gas_fit, 
-                                                        initial_params, unknowns, n), 
+        try
+            func = TwiceDifferentiable(psi_tilde -> log_lik(psi_tilde, y, gas_fit,
+                                                        initial_params, unknowns, n),
                                                         opt_method.initial_points[i])
             opt_result = optimize(func, opt_method, verbose, i, time_limit_sec)
             update_aux_estimation!(aux_est, func, opt_result)
@@ -141,7 +141,7 @@ function fit!(gas::Model{D, T}, y::Vector{T};
         end
     end
 
-    if isempty(aux_est.loglikelihood) 
+    if isempty(aux_est.loglikelihood)
         verbose >= 1 && println("No initial point converged.")
         return
     end
@@ -162,9 +162,9 @@ function fit!(gas::Model{D, T}, y::Vector{T};
     fill_psitilde!(gas, coefs, unknowns)
 
     # Calculate pearson residuals
-    pearson_res = isnan(initial_params[1]) ? 
+    pearson_res = isnan(initial_params[1]) ?
                 quantile_residuals(y, gas) :
                 quantile_residuals(y, gas; initial_params = initial_params)
 
     return Fitted{D, T}(n, unknowns, aic, bic, best_llk, coefs, num_hessian, pearson_res)
-end
+end

src/cross_validation.jl

@@ -38,7 +38,7 @@ end
 """
 TODO
 """
-function cross_validation(gas::Model{<:Distribution, T}, y::Vector{T}, steps_ahead::Int, start_idx::Int;
+function cross_validation(gas::ScoreDrivenModel{<:Distribution, T}, y::Vector{T}, steps_ahead::Int, start_idx::Int;
                   S::Int = 10_000,
                   initial_params::Matrix{T} = DEFAULT_INITIAL_PARAM,
                   opt_method = NelderMead(gas, DEFAULT_NUM_SEEDS)) where T
@@ -50,7 +50,7 @@ function cross_validation(gas::Model{<:Distribution, T}, y::Vector{T}, steps_ahe
         y_to_fit = y[1:start_idx - 1 + i]
         y_to_verify = y[start_idx + i:start_idx - 1 + i + steps_ahead]
         fit!(gas_to_fit, y_to_fit; initial_params=initial_params, opt_method=opt_method)
-        forec = initial_params !== DEFAULT_INITIAL_PARAM ? 
+        forec = initial_params !== DEFAULT_INITIAL_PARAM ?
                     forecast(y_to_fit, gas_to_fit, steps_ahead; S=S, initial_params=initial_params) :
                     forecast(y_to_fit, gas_to_fit, steps_ahead; S=S)
         abs_errors = evaluate_abs_error(y_to_verify, forec.observation_forecast)
@@ -63,4 +63,4 @@ function cross_validation(gas::Model{<:Distribution, T}, y::Vector{T}, steps_ahe
         b.mean_crps[i] = mean(b.crps_scores[i, :])
     end
     return b
-end
+end