bridgr is designed to simplify the implementation and evaluation of bridge models, which are useful for nowcasting (predicting the present or near-term) and forecasting macroeconomic variables like GDP.
Bridge models are statistical tools that link high-frequency indicators (e.g., monthly industrial production) to low-frequency target variables (e.g., quarterly GDP) by forecasting and aggregating the indicators to match the target’s frequency. They enable timely predictions before the official release of low-frequency data, making them essential for policymakers who need early insights for decision-making.
bridgr supports regular frequency ladders from second up to year, several deterministic and model-based indicator forecasting rules, and multiple aggregation choices including joint expalmon and beta weighting. If a target period contains more high-frequency observations than implied by the current frequency mapping, the package keeps the most recent observations and reports a summarized warning.
Compared with midasr, which is a broad toolkit for estimating, testing, selecting, and forecasting MIDAS regressions, bridgr focuses more narrowly on bridge-style nowcasting workflows with automatic frequency alignment, indicator extension, aggregation, and a simpler end-to-end interface. Compared with midasml, which targets high-dimensional mixed-frequency time-series and panel models with regularization methods such as sparse-group LASSO, bridgr is aimed at lower-dimensional applied forecasting settings where transparent bridge-model specification, interpretation, and standard forecasting methods are the priority.
Installation
From CRAN:
install.packages("bridgr")You can install the development version of bridgr like so:
# install.packages("devtools")
devtools::install_github("marcburri/bridgr")Example
This example estimates a bridge model with uncertainty output:
suppressPackageStartupMessages(library(bridgr))
gdp <- suppressMessages(tsbox::ts_na_omit(tsbox::ts_pc(bridgr::gdp)))
bridge_model <- mf_model(
target = gdp,
indic = baro,
indic_predict = "auto.arima",
indic_aggregators = "mean",
indic_lags = 2,
target_lags = 1,
h = 2,
se = TRUE
)
forecast(bridge_model)
#> Mixed-frequency forecast
#> -----------------------------------
#> Target series: gdp
#> Forecast horizon: 2
#> Uncertainty: prediction intervals from residual resampling
#> Simulation paths: 100
#> -----------------------------------
#> time mean se lower_80 upper_80 lower_95 upper_95
#> 1 2023-01-01 0.808 0.794 -0.215 1.870 -1.005 2.059
#> 2 2023-04-01 0.468 0.836 -0.612 1.748 -1.054 2.420
summary(bridge_model)
#> Mixed-frequency model summary
#> -----------------------------------
#> Target series: gdp
#> Target frequency: quarter
#> Forecast horizon: 2
#> Estimation rows: 72
#> Regressors: baro, baro_lag1, baro_lag2, gdp_lag1
#> -----------------------------------
#> Target equation coefficients:
#> Estimate HAC SE
#> (Intercept) -6.356 1.254
#> baro 0.159 0.031
#> baro_lag1 -0.134 0.049
#> baro_lag2 0.042 0.019
#> gdp_lag1 0.223 0.143
#> -----------------------------------
#> Indicator summary:
#> Frequency Predict Aggregation
#> baro month auto.arima mean
#> -----------------------------------
#> Uncertainty:
#> Coefficient SEs: hac
#> Prediction intervals: residual resampling
#> Simulation paths: 100
#> -----------------------------------With se = TRUE, bridgr reports HAC or Delta-HAC coefficient standard errors and residual-resampling prediction intervals.
If you want data-driven within-period weights instead of a simple deterministic aggregator, you can switch the indicator aggregator to expalmon. The corresponding weights are estimated jointly within the bridge model, and solver_options let you control the optimization.
expalmon_model <- mf_model(
target = gdp,
indic = baro,
indic_predict = "auto.arima",
indic_aggregators = "expalmon",
solver_options = list(seed = 123, n_starts = 3),
h = 1
)
summary(expalmon_model)
#> Mixed-frequency model summary
#> -----------------------------------
#> Target series: gdp
#> Target frequency: quarter
#> Forecast horizon: 1
#> Estimation rows: 75
#> Regressors: baro
#> -----------------------------------
#> Target equation coefficients:
#> Estimate
#> (Intercept) -9.371
#> baro 0.098
#> -----------------------------------
#> Indicator summary:
#> Frequency Predict Aggregation
#> baro month auto.arima expalmon
#> -----------------------------------
#> Estimated parametric aggregation:
#> baro weights: 0.006, 0.994, 0.000
#> baro parameters: -4.914, -10.000
#> -----------------------------------
#> Joint parametric aggregation optimization:
#> Method: L-BFGS-B
#> Objective value: 60.832
#> Convergence code: 0
#> Best start: 1 / 3
#> -----------------------------------Code of Conduct
Please note that the bridgr project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.