MinT needs covariance matrix to be positive definite.

[brunocarlin]

Is there a way to enforce positive definite?

I also ran earlier into the same problem as #42 so I rounded both the forecasts and the residuals and replaced any 0s with 1s.

[jnuvenus]

@brunocarlin Hello, I ran into the same problem, have you solved this problem yet? I rounded both the residuals and replaced any 0s with 0.1s or 1s， but still got this error.

[brunocarlin]

I think our problem was the series full of zeros or some other number, try adding random noise, I will be honest , in my final version I stuck with hts because Mint was too advanced for me

[yasineahmed]

Hello,
was anyone able to resolve this issue?
thanks.

[robjhyndman]

This occurs when you have a constant column in your data set. It is impossible to estimate covariances when one of the series is constant. Add some small random noise to the constant column and it should be ok.

[yasineahmed]

Ah, works fine now after adding random noise via rnorm()
Thank you for your assistance.

[dclengacher]

I added noise to both the forecast and residuals and still get this error.

[medoeje]

I added noise to both the forecast and residuals and still get this error.

Same here.
When I'm using ets or arima forecasting algorithms it works well (after removing constant columns), but when I use hybridModel I always get the error.

[edoardomichielon]

Same error here, this time it occurs even if series is not constant, so any of the solution mentioned helps
Thank you for your assistance.

library(dplyr  , quietly = TRUE)
library(tsibble, quietly = TRUE)
library(fable  , quietly = TRUE)

tsibble::tourism %>%
  filter(State == 'ACT') %>% 
  aggregate_key(State / Region, Trips = sum(Trips)) %>% 
  model(base = fable::SNAIVE(Trips)) %>%
  reconcile(mint_cov = min_trace(base, method = "mint_cov")) %>% 
  forecast(h = 1)
#> Error in `mutate()`:
#> ! Problem while computing `mint_cov = (function (object, ...) ...`.
#> Caused by error:
#> ! min_trace needs covariance matrix to be positive definite.

^{Created on 2022-05-26 by the reprex package (v2.0.1)}

Session info

sessioninfo::session_info()
#> ─ Session info ───────────────────────────────────────────────────────────────
#>  setting  value
#>  version  R version 4.1.3 (2022-03-10)
#>  os       Ubuntu 20.04.4 LTS
#>  system   x86_64, linux-gnu
#>  ui       X11
#>  language (EN)
#>  collate  C.UTF-8
#>  ctype    C.UTF-8
#>  tz       Etc/UTC
#>  date     2022-05-26
#>  pandoc   2.11.4 @ /usr/lib/rstudio-server/bin/pandoc/ (via rmarkdown)
#> 
#> ─ Packages ───────────────────────────────────────────────────────────────────
#>  package        * version    date (UTC) lib source
#>  anytime          0.3.9      2020-08-27 [2] CRAN (R 4.0.4)
#>  assertthat       0.2.1      2019-03-21 [2] CRAN (R 4.0.4)
#>  cli              3.3.0      2022-04-25 [2] CRAN (R 4.1.3)
#>  colorspace       2.0-3      2022-02-21 [2] CRAN (R 4.1.2)
#>  crayon           1.5.0      2022-02-14 [2] CRAN (R 4.1.2)
#>  DBI              1.1.2      2021-12-20 [2] CRAN (R 4.1.2)
#>  digest           0.6.29     2021-12-01 [2] CRAN (R 4.1.2)
#>  distributional   0.3.0      2022-01-05 [2] CRAN (R 4.1.2)
#>  dplyr          * 1.0.8      2022-02-08 [2] CRAN (R 4.1.2)
#>  ellipsis         0.3.2      2021-04-29 [2] CRAN (R 4.1.1)
#>  evaluate         0.15       2022-02-18 [2] CRAN (R 4.1.2)
#>  fable          * 0.3.1      2021-05-16 [2] CRAN (R 4.1.1)
#>  fabletools     * 0.3.2.9000 2022-03-16 [2] Github (tidyverts/fabletools@3003027)
#>  fansi            1.0.2      2022-01-14 [2] CRAN (R 4.1.2)
#>  farver           2.1.0      2021-02-28 [2] CRAN (R 4.0.4)
#>  fastmap          1.1.0      2021-01-25 [2] CRAN (R 4.0.4)
#>  fs               1.5.2      2021-12-08 [2] CRAN (R 4.1.2)
#>  generics         0.1.2      2022-01-31 [2] CRAN (R 4.1.2)
#>  ggplot2          3.3.5      2021-06-25 [2] CRAN (R 4.1.1)
#>  glue             1.6.2      2022-02-24 [2] CRAN (R 4.1.2)
#>  gtable           0.3.0      2019-03-25 [2] CRAN (R 4.0.4)
#>  highr            0.9        2021-04-16 [2] CRAN (R 4.1.1)
#>  htmltools        0.5.2      2021-08-25 [2] CRAN (R 4.1.1)
#>  knitr            1.37       2021-12-16 [2] CRAN (R 4.1.2)
#>  lattice          0.20-45    2021-09-22 [4] CRAN (R 4.1.1)
#>  lifecycle        1.0.1      2021-09-24 [2] CRAN (R 4.1.2)
#>  lubridate        1.8.0      2021-10-07 [2] CRAN (R 4.1.2)
#>  magrittr         2.0.2      2022-01-26 [2] CRAN (R 4.1.2)
#>  Matrix           1.4-1      2022-03-23 [4] CRAN (R 4.1.3)
#>  munsell          0.5.0      2018-06-12 [2] CRAN (R 4.0.4)
#>  pillar           1.7.0      2022-02-01 [2] CRAN (R 4.1.2)
#>  pkgconfig        2.0.3      2019-09-22 [2] CRAN (R 4.0.4)
#>  progressr        0.10.0     2021-12-19 [2] CRAN (R 4.1.2)
#>  purrr            0.3.4      2020-04-17 [2] CRAN (R 4.0.4)
#>  R6               2.5.1      2021-08-19 [2] CRAN (R 4.1.1)
#>  Rcpp             1.0.8.2    2022-03-11 [2] CRAN (R 4.1.2)
#>  reprex           2.0.1      2021-08-05 [2] CRAN (R 4.1.1)
#>  rlang            1.0.2      2022-03-04 [2] CRAN (R 4.1.2)
#>  rmarkdown        2.13       2022-03-10 [2] CRAN (R 4.1.2)
#>  rstudioapi       0.13       2020-11-12 [2] CRAN (R 4.0.4)
#>  scales           1.1.1      2020-05-11 [2] CRAN (R 4.0.4)
#>  sessioninfo      1.2.2      2021-12-06 [2] CRAN (R 4.1.2)
#>  stringi          1.7.6      2021-11-29 [2] CRAN (R 4.1.2)
#>  stringr          1.4.0      2019-02-10 [2] CRAN (R 4.0.4)
#>  tibble           3.1.6      2021-11-07 [2] CRAN (R 4.1.2)
#>  tidyr            1.2.0      2022-02-01 [2] CRAN (R 4.1.2)
#>  tidyselect       1.1.2      2022-02-21 [2] CRAN (R 4.1.2)
#>  tsibble        * 1.1.1      2021-12-03 [2] CRAN (R 4.1.2)
#>  utf8             1.2.2      2021-07-24 [2] CRAN (R 4.1.1)
#>  vctrs            0.3.8      2021-04-29 [2] CRAN (R 4.1.1)
#>  withr            2.5.0      2022-03-03 [2] CRAN (R 4.1.2)
#>  xfun             0.30       2022-03-02 [2] CRAN (R 4.1.2)
#>  yaml             2.3.5      2022-02-21 [2] CRAN (R 4.1.2)
#> 
#>  [1] /home/edoardo/R/x86_64-pc-linux-gnu-library/4.1
#>  [2] /usr/local/lib/R/site-library
#>  [3] /usr/lib/R/site-library
#>  [4] /usr/lib/R/library
#> 
#> ──────────────────────────────────────────────────────────────────────────────

[robjhyndman]

First, this has nothing to do with the hts package. You are using the forecast reconcilation procedures in the fabletools package.

Second, your example consists of a single time series (the ACT), so the reconciliation is degenerate.

Third, SNAIVE is already reconciled since the forecasts are equal to the last year of data, so the reconciliation is degenerate.

That said, we should improve fabletools to prevent the error. @mitchelloharawild

To see it working on a small non-degenerate example, try this:

library(dplyr  , quietly = TRUE)
library(tsibble, quietly = TRUE)
library(fable  , quietly = TRUE)

tsibble::tourism %>%
  filter(State == 'ACT') %>% 
  aggregate_key(State / Region, Trips = sum(Trips)) %>% 
  model(base = fable::SNAIVE(Trips)) %>%
  reconcile(mint_cov = min_trace(base, method = "mint_cov")) %>% 
  forecast(h = 1)