In AlbertoAlmuinha/neuralprophet: 'Neural Prophet' Algorithm

knitr::opts_chunk$set(
  collapse = TRUE,
  message = F, 
  warning = F, 
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "90%",
  fig.align = "center"
)

neuralprophet

The NEW Neural Prophet arrives to R!

# Install Development Version 
remotes::install_github("AlbertoAlmuinha/neuralprophet")

Not on CRAN yet:

#install.packages("neuralprophet")

Neural Prophet Algorithm

Using neural_prophet(), which connects to Python NeuralProphet().

library(neuralprophet)
library(tidymodels)
library(tidyverse)
library(timetk)

m <- m4_monthly %>% filter(id == "M750")

splits <- initial_time_split(m)


model_fit_nprophet <- neural_prophet(
    freq            = "M",
    growth          = "linear",
    trend_reg       = 3,
    learn_rate      = 0.1,
    changepoint_range = 0.8,
    seasonality_mode = "additive"
) %>%
    set_engine("prophet") %>%
    fit(value ~ ., training(splits))

# Forecast with 95% Confidence Interval
modeltime_table(
    model_fit_nprophet
) %>%
    modeltime_calibrate(new_data = testing(splits)) %>%
    modeltime_forecast(
        new_data      = testing(splits),
        actual_data   = m,
        conf_interval = 0.95
    ) %>%
    plot_modeltime_forecast(.interactive = FALSE)

Installation

There are 2 key components to installing neuralprophet:

Download the R-Package, neuralprophet. This installs the R-Bindings, which allows you to interface with NeuralProphet
Set up the Python Environment so neuralprophet can connect to the neuralprophet python package.

Step 1 - Download & Install neuralprophet

Download the latest development version.

remotes::install_github("AlbertoAlmuinha/neuralprophet")

Step 2 - Python Environment Setup

Next, you'll need to set up a Python Environment that contains at a minimum:

neuralprophet==0.2.7
pillow==8.3.0
pandas==1.2.5
numpy
pytorch==1.6

The python environment then needs to be activated with reticulate.

Fast Setup

You can use install_nprophet() to prepare and bind to a python environment containing neuralprophet and the required python packages.

You only need to run this one time, and then you are good to go.
Each time you load neuralprophet, the package will include this environment in it's search process.

# Neural Prophet Installation - Run 1st time
install_nprophet()

Step 3 - Restart R and Try an Example

Restart your R session (if in RStudio, close and re-open). Then try this example.

library(neuralprophet)
library(tidymodels)
library(tidyverse)
library(timetk)

m <- m4_monthly %>% filter(id == "M750")

splits <- initial_time_split(m)


model_fit_nprophet <- neural_prophet(
    freq            = "M",
    growth          = "linear",
    trend_reg       = 3,
    learn_rate      = 0.1,
    changepoint_range = 0.8,
    seasonality_mode = "additive"
) %>%
    set_engine("prophet") %>%
    fit(value ~ ., training(splits))

# Forecast with 95% Confidence Interval
modeltime_table(
    model_fit_nprophet
) %>%
    modeltime_calibrate(new_data = testing(splits)) %>%
    modeltime_forecast(
        new_data      = testing(splits),
        actual_data   = m,
        conf_interval = 0.95
    ) %>%
    plot_modeltime_forecast(.interactive = FALSE)

Troubleshooting Installation

Python Environment setup is always fun. Here are a few recommendations if you run into an issue.

Check to make sure Conda or Miniconda is available using reticulate::conda_version(). If no conda version is returned, then use reticulate::install_miniconda() to install Miniconda (recommended vs full Aniconda). Then (re-)run install_gluonts().
Check if Neural Prophet (Python) is available using reticulate::py_module_available("neuralprophet"). If this returns TRUE, then your installation has succeeded in building the environment, but you may have other issues like missing C++ build tools (next).
Windows 10 error: Microsoft Visual C++ is required. Here are the instructions for installing the C++ tools needed.
Other installation issues. Please file a GitHub issue here.