get_n_obs() does not seem to work for WI dp (because sequenceID==NA?)

[Rafnuss]

Using get_n_obs(), I get really low observation counts (<20) while I have usually several hundreads of pictures per deployement.

Could it be possible that this is due to NA value assigned to sequenceID in the import (https://github.com/inbo/camtraptor/blob/HEAD/R/read_wi.R#L326)? According to the documentation (https://inbo.github.io/camtraptor/reference/get_n_obs.html), this is how the number of observations is counted. I am however surprise that I don't get a count of 1 all the time because all sequenceID are NA. Any idea?

[damianooldoni]

Hi @Rafnuss! Thanks for reporting your issues with the package.
Yes, I think sequenceID = NA is the reason of the low number of observations.
The bulk of the calculation is in these lines:
https://github.com/inbo/camtraptor/blob/main/R/get_n_obs.R#L130-L135

However, as you said, seqeunceID equal to NA should result to n = 1.
See example here below using mica dataset:

library(dplyr)
library(camtraptor)

n_obs1 <- get_n_obs(mica1)
n_obs <- get_n_obs(mica)

n_obs1 %>%
  filter(n > 0)
#> # A tibble: 9 × 3
#>   deploymentID                         scientificName         n
#>   <chr>                                <chr>              <int>
#> 1 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Anas strepera          1
#> 2 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Anas platyrhynchos     1
#> 3 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Castor fiber           1
#> 4 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Mustela putorius       1
#> 5 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Vulpes vulpes          1
#> 6 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Martes foina           1
#> 7 62c200a9-0e03-4495-bcd8-032944f6f5a1 Ardea cinerea          1
#> 8 62c200a9-0e03-4495-bcd8-032944f6f5a1 Ardea                  1
#> 9 7ca633fa-64f8-4cfc-a628-6b0c419056d7 Homo sapiens           1

n_obs %>%
  filter(n > 0)
#> # A tibble: 9 × 3
#>   deploymentID                         scientificName         n
#>   <chr>                                <chr>              <int>
#> 1 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Anas strepera          3
#> 2 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Anas platyrhynchos     4
#> 3 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Castor fiber           1
#> 4 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Mustela putorius       3
#> 5 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Vulpes vulpes          1
#> 6 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Martes foina           1
#> 7 62c200a9-0e03-4495-bcd8-032944f6f5a1 Ardea cinerea          1
#> 8 62c200a9-0e03-4495-bcd8-032944f6f5a1 Ardea                  2
#> 9 7ca633fa-64f8-4cfc-a628-6b0c419056d7 Homo sapiens           1


n_obs1 %>%
  filter(n == 0)
#> # A tibble: 27 × 3
#>    deploymentID                         scientificName         n
#>    <chr>                                <chr>              <int>
#>  1 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Castor fiber           0
#>  2 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Mustela putorius       0
#>  3 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Vulpes vulpes          0
#>  4 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Martes foina           0
#>  5 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Ardea cinerea          0
#>  6 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Ardea                  0
#>  7 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 Homo sapiens           0
#>  8 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Anas strepera          0
#>  9 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Anas platyrhynchos     0
#> 10 577b543a-2cf1-4b23-b6d2-cda7e2eac372 Ardea cinerea          0
#> # … with 17 more rows


# all observations per deployment
n_obs_all <- get_n_obs(mica, species = NULL)
n_obs_all
#> # A tibble: 4 × 2
#>   deploymentID                             n
#>   <chr>                                <int>
#> 1 29b7d356-4bb4-4ec4-b792-2af5cc32efa8    10
#> 2 577b543a-2cf1-4b23-b6d2-cda7e2eac372     9
#> 3 62c200a9-0e03-4495-bcd8-032944f6f5a1     5
#> 4 7ca633fa-64f8-4cfc-a628-6b0c419056d7     3

n_obs_all1 <- get_n_obs(mica1, species = NULL)
n_obs_all1
#> # A tibble: 4 × 2
#>   deploymentID                             n
#>   <chr>                                <int>
#> 1 29b7d356-4bb4-4ec4-b792-2af5cc32efa8     3
#> 2 577b543a-2cf1-4b23-b6d2-cda7e2eac372     5
#> 3 62c200a9-0e03-4495-bcd8-032944f6f5a1     3
#> 4 7ca633fa-64f8-4cfc-a628-6b0c419056d7     2

Created on 2022-09-13 with reprex v2.0.2

So, indeed, you should have some ones, not zeros... Can you provide me a reproducible example?

[Rafnuss]

Yes, that's right, I got n>1 because I was simply adding the number of species (get_n_obs(dp, species = NULL)).

So, I guess we should change https://github.com/inbo/camtraptor/blob/main/R/read_wi.R#L308? But see https://github.com/inbo/camtraptor/blob/main/R/read_wi.R#L145. Here is the doc on sequence vs image type: https://www.wildlifeinsights.org/get-started/basics/projects#type.

I'm not really familiar with this differences and their implication for analysis. But intuitively, I would want to be able to define sequenceInterval outside my data (during analysis) and I would expect that a sequence would be computed "on the fly" taking the max count for each species seperatly during that duration.

What would be your recommendation?

[peterdesmet]

I recommend updating read_wi() and assign sequenceID = imageID for image-level datasets.

[damianooldoni]

@Rafnuss: is this solved now?

[Rafnuss]

Sorry, I didn't finished my work last time, here is the fully functional solution (ie. tested) #174

[damianooldoni]

@Rafnuss: is this issue solved?

[PietrH]

Is this solved? Or do we need more tests for #174?

[PietrH]

@damianooldoni Is it ok to close this?