healthyverse_tsa
Time Series Analysis, Modeling and Forecasting of the Healthyverse Packages
Steven P. Sanderson II, MPH - Date: 2026-09-09
Introduction
This analysis follows a Nested Modeltime Workflow from modeltime
along with using the NNS package. I use this to monitor the
downloads of all of my packages:
Get Data
glimpse(downloads_tbl)
Rows: 188,302
Columns: 11
$ date <date> 2020-11-23, 2020-11-23, 2020-11-23, 2020-11-23, 2020-11-23,…
$ time <Period> 15H 36M 55S, 11H 26M 39S, 23H 34M 44S, 18H 39M 32S, 9H 0M…
$ date_time <dttm> 2020-11-23 15:36:55, 2020-11-23 11:26:39, 2020-11-23 23:34:…
$ size <int> 4858294, 4858294, 4858301, 4858295, 361, 4863722, 4864794, 4…
$ r_version <chr> NA, "4.0.3", "3.5.3", "3.5.2", NA, NA, NA, NA, NA, NA, NA, N…
$ r_arch <chr> NA, "x86_64", "x86_64", "x86_64", NA, NA, NA, NA, NA, NA, NA…
$ r_os <chr> NA, "mingw32", "mingw32", "linux-gnu", NA, NA, NA, NA, NA, N…
$ package <chr> "healthyR.data", "healthyR.data", "healthyR.data", "healthyR…
$ version <chr> "1.0.0", "1.0.0", "1.0.0", "1.0.0", "1.0.0", "1.0.0", "1.0.0…
$ country <chr> "US", "US", "US", "GB", "US", "US", "DE", "HK", "JP", "US", …
$ ip_id <int> 2069, 2804, 78827, 27595, 90474, 90474, 42435, 74, 7655, 638…
The last day in the data set is 2026-09-07 23:58:13, the file was birthed on: 2025-10-31 10:47:59.603742, and at report knit time is 7473.17 hours old. Happy analyzing!
Now that we have our data lets take a look at it using the skimr
package.
skim(downloads_tbl)
| Name | downloads_tbl |
| Number of rows | 188302 |
| Number of columns | 11 |
| _______________________ | |
| Column type frequency: | |
| character | 6 |
| Date | 1 |
| numeric | 2 |
| POSIXct | 1 |
| Timespan | 1 |
| ________________________ | |
| Group variables | None |
Data summary
Variable type: character
| skim_variable | n_missing | complete_rate | min | max | empty | n_unique | whitespace |
|---|---|---|---|---|---|---|---|
| r_version | 141536 | 0.25 | 5 | 17 | 0 | 54 | 0 |
| r_arch | 141536 | 0.25 | 1 | 7 | 0 | 7 | 0 |
| r_os | 141536 | 0.25 | 7 | 33 | 0 | 39 | 0 |
| package | 0 | 1.00 | 7 | 13 | 0 | 8 | 0 |
| version | 0 | 1.00 | 5 | 17 | 0 | 63 | 0 |
| country | 18390 | 0.90 | 2 | 2 | 0 | 172 | 0 |
Variable type: Date
| skim_variable | n_missing | complete_rate | min | max | median | n_unique |
|---|---|---|---|---|---|---|
| date | 0 | 1 | 2020-11-23 | 2026-09-07 | 2024-03-06 | 2108 |
Variable type: numeric
| skim_variable | n_missing | complete_rate | mean | sd | p0 | p25 | p50 | p75 | p100 | hist |
|---|---|---|---|---|---|---|---|---|---|---|
| size | 0 | 1 | 1140931.93 | 1472136.49 | 355 | 46550 | 329202.5 | 2354658 | 5677952 | ▇▁▂▁▁ |
| ip_id | 0 | 1 | 12561.65 | 26412.84 | 1 | 163 | 2742.0 | 12327 | 429286 | ▇▁▁▁▁ |
Variable type: POSIXct
| skim_variable | n_missing | complete_rate | min | max | median | n_unique |
|---|---|---|---|---|---|---|
| date_time | 0 | 1 | 2020-11-23 09:00:41 | 2026-09-07 23:58:13 | 2024-03-06 04:45:33 | 121661 |
Variable type: Timespan
| skim_variable | n_missing | complete_rate | min | max | median | n_unique |
|---|---|---|---|---|---|---|
| time | 0 | 1 | 0 | 59 | 32 | 60 |
We can see that the following columns are missing a lot of data and for
us are most likely not useful anyways, so we will drop them
c(r_version, r_arch, r_os)
Plots
Now lets take a look at a time-series plot of the total daily downloads by package. We will use a log scale and place a vertical line at each version release for each package.
[[1]]
[[2]]
[[3]]
[[4]]
[[5]]
[[6]]
[[7]]
[[8]]
Now lets take a look at some time series decomposition graphs.
[[1]]
[[2]]
[[3]]
[[4]]
[[5]]
[[6]]
[[7]]
[[8]]
[[1]]
[[2]]
[[3]]
[[4]]
[[5]]
[[6]]
[[7]]
[[8]]
Seasonal Diagnostics:
[[1]]
[[2]]
[[3]]
[[4]]
[[5]]
[[6]]
[[7]]
[[8]]
ACF and PACF Diagnostics:
[[1]]
[[2]]
[[3]]
[[4]]
[[5]]
[[6]]
[[7]]
[[8]]
Feature Engineering
Now that we have our basic data and a shot of what it looks like, let’s
add some features to our data which can be very helpful in modeling.
Lets start by making a tibble that is aggregated by the day and
package, as we are going to be interested in forecasting the next 4
weeks or 28 days for each package. First lets get our base data.
Call:
stats::lm(formula = .formula, data = df)
Residuals:
Min 1Q Median 3Q Max
-151.55 -38.67 -11.85 28.30 826.55
Coefficients:
Estimate Std. Error
(Intercept) -1.229e+02 4.720e+01
date 8.176e-03 2.485e-03
lag(value, 1) 9.876e-02 2.171e-02
lag(value, 7) 7.125e-02 2.234e-02
lag(value, 14) 7.680e-02 2.215e-02
lag(value, 21) 8.801e-02 2.224e-02
lag(value, 28) 8.335e-02 2.216e-02
lag(value, 35) 3.507e-02 2.219e-02
lag(value, 42) 6.115e-02 2.230e-02
lag(value, 49) 7.258e-02 2.225e-02
month(date, label = TRUE).L -8.480e+00 4.726e+00
month(date, label = TRUE).Q 2.020e+00 4.617e+00
month(date, label = TRUE).C -1.501e+01 4.676e+00
month(date, label = TRUE)^4 -9.871e+00 4.711e+00
month(date, label = TRUE)^5 -5.391e+00 4.660e+00
month(date, label = TRUE)^6 4.895e-01 4.694e+00
month(date, label = TRUE)^7 -2.033e+00 4.623e+00
month(date, label = TRUE)^8 -4.298e+00 4.598e+00
month(date, label = TRUE)^9 -3.543e-01 4.605e+00
month(date, label = TRUE)^10 -3.412e-01 4.544e+00
month(date, label = TRUE)^11 -5.152e-01 4.477e+00
fourier_vec(date, type = "sin", K = 1, period = 7) -1.031e+01 2.056e+00
fourier_vec(date, type = "cos", K = 1, period = 7) 6.946e+00 2.114e+00
t value Pr(>|t|)
(Intercept) -2.604 0.009275 **
date 3.290 0.001018 **
lag(value, 1) 4.549 5.71e-06 ***
lag(value, 7) 3.189 0.001447 **
lag(value, 14) 3.468 0.000536 ***
lag(value, 21) 3.958 7.81e-05 ***
lag(value, 28) 3.761 0.000174 ***
lag(value, 35) 1.580 0.114224
lag(value, 42) 2.742 0.006157 **
lag(value, 49) 3.262 0.001123 **
month(date, label = TRUE).L -1.794 0.072899 .
month(date, label = TRUE).Q 0.438 0.661766
month(date, label = TRUE).C -3.211 0.001345 **
month(date, label = TRUE)^4 -2.096 0.036243 *
month(date, label = TRUE)^5 -1.157 0.247444
month(date, label = TRUE)^6 0.104 0.916951
month(date, label = TRUE)^7 -0.440 0.660175
month(date, label = TRUE)^8 -0.935 0.350058
month(date, label = TRUE)^9 -0.077 0.938689
month(date, label = TRUE)^10 -0.075 0.940146
month(date, label = TRUE)^11 -0.115 0.908400
fourier_vec(date, type = "sin", K = 1, period = 7) -5.013 5.82e-07 ***
fourier_vec(date, type = "cos", K = 1, period = 7) 3.286 0.001032 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 60.14 on 2036 degrees of freedom
(49 observations deleted due to missingness)
Multiple R-squared: 0.1946, Adjusted R-squared: 0.1859
F-statistic: 22.36 on 22 and 2036 DF, p-value: < 2.2e-16
NNS Forecasting
This is something I have been wanting to try for a while. The NNS
package is a great package for forecasting time series data.
library(NNS)
data_list <- base_data |>
select(package, value) |>
group_split(package)
data_list |>
imap(
\(x, idx) {
obj <- x
x <- obj |> pull(value) |> tail(7*52)
train_set_size <- length(x) - 56
pkg <- obj |> pluck(1) |> unique()
# sf <- NNS.seas(x, modulo = 7, plot = FALSE)$periods
seas <- t(
sapply(
1:25,
function(i) c(
i,
sqrt(
mean((
NNS.ARMA(x,
h = 28,
training.set = train_set_size,
method = "lin",
seasonal.factor = i,
plot=FALSE
) - tail(x, 28)) ^ 2)))
)
)
colnames(seas) <- c("Period", "RMSE")
sf <- seas[which.min(seas[, 2]), 1]
cat(paste0("Package: ", pkg, "\n"))
NNS.ARMA.optim(
variable = x,
h = 28,
training.set = train_set_size,
#seasonal.factor = seq(12, 60, 7),
seasonal.factor = sf,
pred.int = 0.95,
plot = TRUE
)
title(
sub = paste0("\n",
"Package: ", pkg, " - NNS Optimization")
)
}
)
Package: healthyR
[1] "CURRNET METHOD: lin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'lin' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT lin OBJECTIVE FUNCTION = 5.6043077477447"
[1] "BEST method = 'lin' PATH MEMBER = c( 19 )"
[1] "BEST lin OBJECTIVE FUNCTION = 5.6043077477447"
[1] "CURRNET METHOD: nonlin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'nonlin' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT nonlin OBJECTIVE FUNCTION = 4.94273363775119"
[1] "BEST method = 'nonlin' PATH MEMBER = c( 19 )"
[1] "BEST nonlin OBJECTIVE FUNCTION = 4.94273363775119"
[1] "CURRNET METHOD: both"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'both' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT both OBJECTIVE FUNCTION = 4.2522389573637"
[1] "BEST method = 'both' PATH MEMBER = c( 19 )"
[1] "BEST both OBJECTIVE FUNCTION = 4.2522389573637"
Package: healthyR.ai
[1] "CURRNET METHOD: lin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'lin' , seasonal.factor = c( 22 ) ...)"
[1] "CURRENT lin OBJECTIVE FUNCTION = 6.36203963059885"
[1] "BEST method = 'lin' PATH MEMBER = c( 22 )"
[1] "BEST lin OBJECTIVE FUNCTION = 6.36203963059885"
[1] "CURRNET METHOD: nonlin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'nonlin' , seasonal.factor = c( 22 ) ...)"
[1] "CURRENT nonlin OBJECTIVE FUNCTION = 6.37599749998468"
[1] "BEST method = 'nonlin' PATH MEMBER = c( 22 )"
[1] "BEST nonlin OBJECTIVE FUNCTION = 6.37599749998468"
[1] "CURRNET METHOD: both"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'both' , seasonal.factor = c( 22 ) ...)"
[1] "CURRENT both OBJECTIVE FUNCTION = 6.35900726255883"
[1] "BEST method = 'both' PATH MEMBER = c( 22 )"
[1] "BEST both OBJECTIVE FUNCTION = 6.35900726255883"
Package: healthyR.data
[1] "CURRNET METHOD: lin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'lin' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT lin OBJECTIVE FUNCTION = 11.3802946592186"
[1] "BEST method = 'lin' PATH MEMBER = c( 19 )"
[1] "BEST lin OBJECTIVE FUNCTION = 11.3802946592186"
[1] "CURRNET METHOD: nonlin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'nonlin' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT nonlin OBJECTIVE FUNCTION = 10.8166042094465"
[1] "BEST method = 'nonlin' PATH MEMBER = c( 19 )"
[1] "BEST nonlin OBJECTIVE FUNCTION = 10.8166042094465"
[1] "CURRNET METHOD: both"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'both' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT both OBJECTIVE FUNCTION = 10.1135641909011"
[1] "BEST method = 'both' PATH MEMBER = c( 19 )"
[1] "BEST both OBJECTIVE FUNCTION = 10.1135641909011"
Package: healthyR.ts
[1] "CURRNET METHOD: lin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'lin' , seasonal.factor = c( 14 ) ...)"
[1] "CURRENT lin OBJECTIVE FUNCTION = 5.12069665302342"
[1] "BEST method = 'lin' PATH MEMBER = c( 14 )"
[1] "BEST lin OBJECTIVE FUNCTION = 5.12069665302342"
[1] "CURRNET METHOD: nonlin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'nonlin' , seasonal.factor = c( 14 ) ...)"
[1] "CURRENT nonlin OBJECTIVE FUNCTION = 9.00833094363996"
[1] "BEST method = 'nonlin' PATH MEMBER = c( 14 )"
[1] "BEST nonlin OBJECTIVE FUNCTION = 9.00833094363996"
[1] "CURRNET METHOD: both"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'both' , seasonal.factor = c( 14 ) ...)"
[1] "CURRENT both OBJECTIVE FUNCTION = 6.08736951572171"
[1] "BEST method = 'both' PATH MEMBER = c( 14 )"
[1] "BEST both OBJECTIVE FUNCTION = 6.08736951572171"
Package: healthyverse
[1] "CURRNET METHOD: lin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'lin' , seasonal.factor = c( 5 ) ...)"
[1] "CURRENT lin OBJECTIVE FUNCTION = 17.1164541932593"
[1] "BEST method = 'lin' PATH MEMBER = c( 5 )"
[1] "BEST lin OBJECTIVE FUNCTION = 17.1164541932593"
[1] "CURRNET METHOD: nonlin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'nonlin' , seasonal.factor = c( 5 ) ...)"
[1] "CURRENT nonlin OBJECTIVE FUNCTION = 13.9036425865023"
[1] "BEST method = 'nonlin' PATH MEMBER = c( 5 )"
[1] "BEST nonlin OBJECTIVE FUNCTION = 13.9036425865023"
[1] "CURRNET METHOD: both"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'both' , seasonal.factor = c( 5 ) ...)"
[1] "CURRENT both OBJECTIVE FUNCTION = 18.6446423268379"
[1] "BEST method = 'both' PATH MEMBER = c( 5 )"
[1] "BEST both OBJECTIVE FUNCTION = 18.6446423268379"
Package: RandomWalker
[1] "CURRNET METHOD: lin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'lin' , seasonal.factor = c( 14 ) ...)"
[1] "CURRENT lin OBJECTIVE FUNCTION = 5.90151466407107"
[1] "BEST method = 'lin' PATH MEMBER = c( 14 )"
[1] "BEST lin OBJECTIVE FUNCTION = 5.90151466407107"
[1] "CURRNET METHOD: nonlin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'nonlin' , seasonal.factor = c( 14 ) ...)"
[1] "CURRENT nonlin OBJECTIVE FUNCTION = 24.8833145964153"
[1] "BEST method = 'nonlin' PATH MEMBER = c( 14 )"
[1] "BEST nonlin OBJECTIVE FUNCTION = 24.8833145964153"
[1] "CURRNET METHOD: both"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'both' , seasonal.factor = c( 14 ) ...)"
[1] "CURRENT both OBJECTIVE FUNCTION = 17.9829248381469"
[1] "BEST method = 'both' PATH MEMBER = c( 14 )"
[1] "BEST both OBJECTIVE FUNCTION = 17.9829248381469"
Package: tidyAML
[1] "CURRNET METHOD: lin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'lin' , seasonal.factor = c( 25 ) ...)"
[1] "CURRENT lin OBJECTIVE FUNCTION = 11.2934696216867"
[1] "BEST method = 'lin' PATH MEMBER = c( 25 )"
[1] "BEST lin OBJECTIVE FUNCTION = 11.2934696216867"
[1] "CURRNET METHOD: nonlin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'nonlin' , seasonal.factor = c( 25 ) ...)"
[1] "CURRENT nonlin OBJECTIVE FUNCTION = 8.04321308691866"
[1] "BEST method = 'nonlin' PATH MEMBER = c( 25 )"
[1] "BEST nonlin OBJECTIVE FUNCTION = 8.04321308691866"
[1] "CURRNET METHOD: both"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'both' , seasonal.factor = c( 25 ) ...)"
[1] "CURRENT both OBJECTIVE FUNCTION = 7.78936462099608"
[1] "BEST method = 'both' PATH MEMBER = c( 25 )"
[1] "BEST both OBJECTIVE FUNCTION = 7.78936462099608"
Package: TidyDensity
[1] "CURRNET METHOD: lin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'lin' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT lin OBJECTIVE FUNCTION = 5.59232011480999"
[1] "BEST method = 'lin' PATH MEMBER = c( 19 )"
[1] "BEST lin OBJECTIVE FUNCTION = 5.59232011480999"
[1] "CURRNET METHOD: nonlin"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'nonlin' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT nonlin OBJECTIVE FUNCTION = 6.0218719925543"
[1] "BEST method = 'nonlin' PATH MEMBER = c( 19 )"
[1] "BEST nonlin OBJECTIVE FUNCTION = 6.0218719925543"
[1] "CURRNET METHOD: both"
[1] "COPY LATEST PARAMETERS DIRECTLY FOR NNS.ARMA() IF ERROR:"
[1] "NNS.ARMA(... method = 'both' , seasonal.factor = c( 19 ) ...)"
[1] "CURRENT both OBJECTIVE FUNCTION = 5.06923395530877"
[1] "BEST method = 'both' PATH MEMBER = c( 19 )"
[1] "BEST both OBJECTIVE FUNCTION = 5.06923395530877"
[[1]]
NULL
[[2]]
NULL
[[3]]
NULL
[[4]]
NULL
[[5]]
NULL
[[6]]
NULL
[[7]]
NULL
[[8]]
NULL
Pre-Processing
Now we are going to do some basic pre-processing.
data_padded_tbl <- base_data %>%
pad_by_time(
.date_var = date,
.pad_value = 0
)
# Get log interval and standardization parameters
log_params <- liv(data_padded_tbl$value, limit_lower = 0, offset = 1, silent = TRUE)
limit_lower <- log_params$limit_lower
limit_upper <- log_params$limit_upper
offset <- log_params$offset
data_liv_tbl <- data_padded_tbl %>%
# Get log interval transform
mutate(value_trans = liv(value, limit_lower = 0, offset = 1, silent = TRUE)$log_scaled)
# Get Standardization Params
std_params <- standard_vec(data_liv_tbl$value_trans, silent = TRUE)
std_mean <- std_params$mean
std_sd <- std_params$sd
data_transformed_tbl <- data_liv_tbl %>%
group_by(package) %>%
# get standardization
mutate(value_trans = standard_vec(value_trans, silent = TRUE)$standard_scaled) %>%
tk_augment_fourier(
.date_var = date,
.periods = c(7, 14, 30, 90, 180),
.K = 2
) %>%
tk_augment_timeseries_signature(
.date_var = date
) %>%
ungroup() %>%
select(-c(value, -year.iso))
Since this is panel data we can follow one of two different modeling strategies. We can search for a global model in the panel data or we can use nested forecasting finding the best model for each of the time series. Since we only have 5 panels, we will use nested forecasting.
To do this we will use the nest_timeseries and
split_nested_timeseries functions to create a nested tibble.
horizon <- 4*7
nested_data_tbl <- data_transformed_tbl %>%
# 0. Filter out column where package is NA
filter(!is.na(package)) %>%
# 1. Extending: We'll predict n days into the future.
extend_timeseries(
.id_var = package,
.date_var = date,
.length_future = horizon
) %>%
# 2. Nesting: We'll group by id, and create a future dataset
# that forecasts n days of extended data and
# an actual dataset that contains n*2 days
nest_timeseries(
.id_var = package,
.length_future = horizon
#.length_actual = horizon*2
) %>%
# 3. Splitting: We'll take the actual data and create splits
# for accuracy and confidence interval estimation of n das (test)
# and the rest is training data
split_nested_timeseries(
.length_test = horizon
)
nested_data_tbl
# A tibble: 8 × 4
package .actual_data .future_data .splits
<fct> <list> <list> <list>
1 healthyR.data <tibble [2,096 × 50]> <tibble [28 × 50]> <split [2068|28]>
2 healthyR <tibble [2,090 × 50]> <tibble [28 × 50]> <split [2062|28]>
3 healthyR.ts <tibble [2,026 × 50]> <tibble [28 × 50]> <split [1998|28]>
4 healthyverse <tibble [1,905 × 50]> <tibble [28 × 50]> <split [1877|28]>
5 healthyR.ai <tibble [1,831 × 50]> <tibble [28 × 50]> <split [1803|28]>
6 TidyDensity <tibble [1,684 × 50]> <tibble [28 × 50]> <split [1656|28]>
7 tidyAML <tibble [1,287 × 50]> <tibble [28 × 50]> <split [1259|28]>
8 RandomWalker <tibble [711 × 50]> <tibble [28 × 50]> <split [683|28]>
Now it is time to make some recipes and models using the modeltime workflow.
Modeltime Workflow
Recipe Object
recipe_base <- recipe(
value_trans ~ .
, data = extract_nested_test_split(nested_data_tbl)
)
recipe_base
recipe_date <- recipe(
value_trans ~ date
, data = extract_nested_test_split(nested_data_tbl)
)
Models
# Models ------------------------------------------------------------------
# Auto ARIMA --------------------------------------------------------------
model_spec_arima_no_boost <- arima_reg() %>%
set_engine(engine = "auto_arima")
wflw_auto_arima <- workflow() %>%
add_recipe(recipe = recipe_date) %>%
add_model(model_spec_arima_no_boost)
# NNETAR ------------------------------------------------------------------
model_spec_nnetar <- nnetar_reg(
mode = "regression"
, seasonal_period = "auto"
) %>%
set_engine("nnetar")
wflw_nnetar <- workflow() %>%
add_recipe(recipe = recipe_base) %>%
add_model(model_spec_nnetar)
# TSLM --------------------------------------------------------------------
model_spec_lm <- linear_reg() %>%
set_engine("lm")
wflw_lm <- workflow() %>%
add_recipe(recipe = recipe_base) %>%
add_model(model_spec_lm)
# MARS --------------------------------------------------------------------
model_spec_mars <- mars(mode = "regression") %>%
set_engine("earth")
wflw_mars <- workflow() %>%
add_recipe(recipe = recipe_date) %>%
add_model(model_spec_mars)
Nested Modeltime Tables
nested_modeltime_tbl <- modeltime_nested_fit(
# Nested Data
nested_data = nested_data_tbl,
control = control_nested_fit(
verbose = TRUE,
allow_par = FALSE
),
# Add workflows
wflw_auto_arima,
wflw_lm,
wflw_mars,
wflw_nnetar
)
nested_modeltime_tbl <- nested_modeltime_tbl[!is.na(nested_modeltime_tbl$package),]
Model Accuracy
nested_modeltime_tbl %>%
extract_nested_test_accuracy() %>%
filter(!is.na(package)) %>%
knitr::kable()
| package | .model_id | .model_desc | .type | mae | mape | mase | smape | rmse | rsq |
|---|---|---|---|---|---|---|---|---|---|
| healthyR.data | 1 | ARIMA | Test | 0.8214067 | 423.3073 | 0.6293988 | 122.21771 | 1.0719371 | 0.0079288 |
| healthyR.data | 2 | LM | Test | 0.7912731 | 529.8779 | 0.6063090 | 122.30888 | 1.0457989 | 0.0087175 |
| healthyR.data | 3 | EARTH | Test | 5.5353115 | 4339.6429 | 4.2414047 | 182.55327 | 6.0303505 | 0.0912142 |
| healthyR.data | 4 | NNAR | Test | 0.8029431 | 518.4518 | 0.6152511 | 126.60729 | 1.0404334 | 0.0220981 |
| healthyR | 1 | ARIMA | Test | 0.7570316 | 301.2617 | 0.7319784 | 127.74055 | 0.9329499 | 0.0181410 |
| healthyR | 2 | LM | Test | 0.8044477 | 260.3289 | 0.7778254 | 131.55618 | 1.0437135 | 0.0109078 |
| healthyR | 3 | EARTH | Test | 2.5077737 | 1476.9466 | 2.4247817 | 175.41243 | 2.7969593 | 0.1653644 |
| healthyR | 4 | NNAR | Test | 0.6615984 | 140.6529 | 0.6397035 | 133.92025 | 0.8775111 | 0.1010255 |
| healthyR.ts | 1 | ARIMA | Test | 0.5645618 | 331.8148 | 0.6829879 | 111.31123 | 0.7528614 | 0.0082659 |
| healthyR.ts | 2 | LM | Test | 0.6936134 | 422.4811 | 0.8391101 | 168.67772 | 0.7965853 | 0.0006473 |
| healthyR.ts | 3 | EARTH | Test | 0.7128802 | 480.6145 | 0.8624184 | 118.64584 | 0.8770672 | 0.0625081 |
| healthyR.ts | 4 | NNAR | Test | 0.6655983 | 354.2060 | 0.8052184 | 153.85452 | 0.7973727 | 0.0006676 |
| healthyverse | 1 | ARIMA | Test | 0.5147528 | 258.8732 | 0.9238375 | 44.56867 | 0.6250428 | 0.0143777 |
| healthyverse | 2 | LM | Test | 0.7861959 | 216.7522 | 1.4110020 | 78.51723 | 0.9226040 | 0.0100059 |
| healthyverse | 3 | EARTH | Test | 0.4958259 | 303.9026 | 0.8898690 | 41.58116 | 0.6708712 | 0.0385566 |
| healthyverse | 4 | NNAR | Test | 0.9411170 | 204.7580 | 1.6890422 | 100.99996 | 1.0410958 | 0.0012186 |
| healthyR.ai | 1 | ARIMA | Test | 0.8746560 | 152.8643 | 1.0795743 | 145.69247 | 1.0328594 | 0.1090918 |
| healthyR.ai | 2 | LM | Test | 0.9445083 | 157.3660 | 1.1657920 | 157.20279 | 1.1051544 | 0.0444481 |
| healthyR.ai | 3 | EARTH | Test | 2.8984405 | 670.7082 | 3.5775003 | 160.47196 | 3.1789482 | 0.2855280 |
| healthyR.ai | 4 | NNAR | Test | 0.9561294 | 154.1105 | 1.1801357 | 156.31372 | 1.1297796 | 0.0077178 |
| TidyDensity | 1 | ARIMA | Test | 0.9102311 | 235.5154 | 0.8346333 | 142.81712 | 1.0984217 | 0.0316294 |
| TidyDensity | 2 | LM | Test | 0.7701492 | 198.7797 | 0.7061857 | 142.54863 | 0.8783090 | 0.0221856 |
| TidyDensity | 3 | EARTH | Test | 2.7508609 | 1250.4174 | 2.5223925 | 153.43564 | 3.1076936 | 0.0616347 |
| TidyDensity | 4 | NNAR | Test | 0.7661659 | 232.6606 | 0.7025332 | 137.17611 | 0.8905447 | 0.0455362 |
| tidyAML | 1 | ARIMA | Test | 0.9414153 | 118.9509 | 1.3447336 | 183.26298 | 1.0932761 | 0.0017669 |
| tidyAML | 2 | LM | Test | 1.0175523 | 119.3348 | 1.4534890 | 160.14704 | 1.2160093 | 0.0044608 |
| tidyAML | 3 | EARTH | Test | 0.9333616 | 101.6304 | 1.3332296 | 187.60134 | 1.0909329 | 0.2252899 |
| tidyAML | 4 | NNAR | Test | 1.0399672 | 150.6841 | 1.4855068 | 154.93084 | 1.2692850 | 0.0218667 |
| RandomWalker | 1 | ARIMA | Test | 0.7870756 | 249.1851 | 0.7810382 | 122.63982 | 0.9550054 | 0.1824893 |
| RandomWalker | 2 | LM | Test | 0.6062261 | 108.2093 | 0.6015759 | 160.69376 | 0.6940545 | 0.0250432 |
| RandomWalker | 3 | EARTH | Test | 1.3250794 | 406.7844 | 1.3149153 | 137.08975 | 1.5055904 | 0.0059858 |
| RandomWalker | 4 | NNAR | Test | 0.6298964 | 127.0596 | 0.6250647 | 140.38101 | 0.7670723 | 0.0000535 |
Plot Models
nested_modeltime_tbl %>%
extract_nested_test_forecast() %>%
group_by(package) %>%
filter_by_time(.date_var = .index, .start_date = max(.index) - 60) %>%
ungroup() %>%
plot_modeltime_forecast(
.interactive = FALSE,
.conf_interval_show = FALSE,
.facet_scales = "free"
) +
theme_minimal() +
facet_wrap(~ package, nrow = 3) +
theme(legend.position = "bottom")
Best Model
best_nested_modeltime_tbl <- nested_modeltime_tbl %>%
modeltime_nested_select_best(
metric = "rmse",
minimize = TRUE,
filter_test_forecasts = TRUE
)
best_nested_modeltime_tbl %>%
extract_nested_best_model_report()
# Nested Modeltime Table
# A tibble: 8 × 10
package .model_id .model_desc .type mae mape mase smape rmse rsq
<fct> <int> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 healthyR.da… 4 NNAR Test 0.803 518. 0.615 127. 1.04 0.0221
2 healthyR 4 NNAR Test 0.662 141. 0.640 134. 0.878 0.101
3 healthyR.ts 1 ARIMA Test 0.565 332. 0.683 111. 0.753 0.00827
4 healthyverse 1 ARIMA Test 0.515 259. 0.924 44.6 0.625 0.0144
5 healthyR.ai 1 ARIMA Test 0.875 153. 1.08 146. 1.03 0.109
6 TidyDensity 2 LM Test 0.770 199. 0.706 143. 0.878 0.0222
7 tidyAML 3 EARTH Test 0.933 102. 1.33 188. 1.09 0.225
8 RandomWalker 2 LM Test 0.606 108. 0.602 161. 0.694 0.0250
best_nested_modeltime_tbl %>%
extract_nested_test_forecast() %>%
#filter(!is.na(.model_id)) %>%
group_by(package) %>%
filter_by_time(.date_var = .index, .start_date = max(.index) - 60) %>%
ungroup() %>%
plot_modeltime_forecast(
.interactive = FALSE,
.conf_interval_alpha = 0.2,
.facet_scales = "free"
) +
facet_wrap(~ package, nrow = 3) +
theme_minimal() +
theme(legend.position = "bottom")
Refitting and Future Forecast
Now that we have the best models, we can make our future forecasts.
nested_modeltime_refit_tbl <- best_nested_modeltime_tbl %>%
modeltime_nested_refit(
control = control_nested_refit(verbose = TRUE)
)
nested_modeltime_refit_tbl
# Nested Modeltime Table
# A tibble: 8 × 5
package .actual_data .future_data .splits .modeltime_tables
<fct> <list> <list> <list> <list>
1 healthyR.data <tibble> <tibble> <split [2068|28]> <mdl_tm_t [1 × 5]>
2 healthyR <tibble> <tibble> <split [2062|28]> <mdl_tm_t [1 × 5]>
3 healthyR.ts <tibble> <tibble> <split [1998|28]> <mdl_tm_t [1 × 5]>
4 healthyverse <tibble> <tibble> <split [1877|28]> <mdl_tm_t [1 × 5]>
5 healthyR.ai <tibble> <tibble> <split [1803|28]> <mdl_tm_t [1 × 5]>
6 TidyDensity <tibble> <tibble> <split [1656|28]> <mdl_tm_t [1 × 5]>
7 tidyAML <tibble> <tibble> <split [1259|28]> <mdl_tm_t [1 × 5]>
8 RandomWalker <tibble> <tibble> <split [683|28]> <mdl_tm_t [1 × 5]>
nested_modeltime_refit_tbl %>%
extract_nested_future_forecast() %>%
group_by(package) %>%
mutate(across(.value:.conf_hi, .fns = ~ standard_inv_vec(
x = .,
mean = std_mean,
sd = std_sd
)$standard_inverse_value)) %>%
mutate(across(.value:.conf_hi, .fns = ~ liiv(
x = .,
limit_lower = limit_lower,
limit_upper = limit_upper,
offset = offset
)$rescaled_v)) %>%
filter_by_time(.date_var = .index, .start_date = max(.index) - 60) %>%
ungroup() %>%
plot_modeltime_forecast(
.interactive = FALSE,
.conf_interval_alpha = 0.2,
.facet_scales = "free"
) +
facet_wrap(~ package, nrow = 3) +
theme_minimal() +
theme(legend.position = "bottom")
