Probabilistic Time Series Forecasting.

ngboostForecast

The goal of ngboostForecast is to provide a tools for probabilistic forecasting by using Python’s ngboost for R users.

Installation

You can install the released version of ngboostForecast from CRAN with:

install.packages("ngboostForecast")

And the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("Akai01/ngboostForecast")

Example

This is a basic example which shows you how to solve a common problem:

library(ngboostForecast)
#> Loading required package: reticulate
#> Registered S3 method overwritten by 'quantmod':
#>   method            from
#>   as.zoo.data.frame zoo

train = window(seatbelts, end = c(1983,12))

test = window(seatbelts, start = c(1984,1))

# without external variables with Ridge regression

model <- NGBforecast$new(Dist = Dist("LogNormal"),
                         Base = sklearner(module = "linear_model",
                                          class = "Ridge"),
                         Score = Scores("LogScore"),
                         natural_gradient = TRUE,
                         n_estimators = 200,
                         learning_rate =  0.1,
                         minibatch_frac = 1,
                         col_sample = 1,
                         verbose = TRUE,
                         verbose_eval = 5,
                         tol = 1e-5)

model$fit(y = train[,2], 
          seasonal = TRUE, 
          max_lag = 12, 
          early_stopping_rounds = 10L)

fc <- model$forecast(h = 12, level = c(99,95,90, 80, 70, 60), 
                      data_frame = FALSE)

autoplot(fc) + autolayer(test[,2])

Tuning

Set the parameters:

library(ngboostForecast)

dists <- list(Dist("Normal"))

base_learners <- list(sklearner(module = "tree", class = "DecisionTreeRegressor",
                                max_depth = 1),
                      sklearner(module = "tree", class = "DecisionTreeRegressor",
                                max_depth = 2),
                      sklearner(module = "tree", class = "DecisionTreeRegressor",
                                max_depth = 3),
                      sklearner(module = "tree", class = "DecisionTreeRegressor",
                                max_depth = 4),
                      sklearner(module = "tree", class = "DecisionTreeRegressor",
                                max_depth = 5),
                      sklearner(module = "tree", class = "DecisionTreeRegressor",
                                max_depth = 6),
                      sklearner(module = "tree", class = "DecisionTreeRegressor",
                                max_depth = 7))

scores <-  list(Scores("LogScore"))

model <- NGBforecastCV$new(Dist = dists,
                           Base = base_learners,
                           Score = scores,
                           natural_gradient = TRUE,
                           n_estimators = list(10, 100),
                           learning_rate = list(0.1, 0.2),
                           minibatch_frac = list(0.1, 1),
                           col_sample = list(0.3),
                           verbose = FALSE,
                           verbose_eval = 100,
                           tol = 1e-5)

Tune the model:

params <- model$tune(y = AirPassengers,
                     seasonal = TRUE,
                     max_lag = 12,
                     xreg = NULL,
                     early_stopping_rounds = NULL,
                     n_splits = 4L)

Best parameters:

params
#> $ngboost_best_params
#> $ngboost_best_params$Base
#> DecisionTreeRegressor(max_depth=7.0)
#> 
#> $ngboost_best_params$Dist
#> <class 'ngboost.distns.normal.Normal'>
#> 
#> $ngboost_best_params$Score
#> <class 'ngboost.scores.LogScore'>
#> 
#> $ngboost_best_params$col_sample
#> [1] 0.3
#> 
#> $ngboost_best_params$learning_rate
#> [1] 0.2
#> 
#> $ngboost_best_params$minibatch_frac
#> [1] 1
#> 
#> $ngboost_best_params$n_estimators
#> [1] 100
#> 
#> 
#> $ngb_forecast_params
#> $ngb_forecast_params$seasonal
#> [1] TRUE
#> 
#> $ngb_forecast_params$max_lag
#> [1] 12
#> 
#> $ngb_forecast_params$K
#> [1] 5