Abel set it up as one row which is not what I have it as.

@SaraDeLessio can you give me the CSS to make the labels inline with the selectors, make the text the same font as the selectors and make the selectors go the full width of the page (perhaps the first selector will look odd full-width) but let's see.

Unless you have a better idea

Add function to compare years

It gives an error when these Qs are selected

@abelvaldivia starting with 61e the else if part of your code does not match your selected variable. Below is an example. You can see else if refers to 61e but the variable selected is 61f. Is this a mistake?

If it's a mistake, there there is one plot missing. 61e is not there since it jumps to 61f.

      else if (input$hhs_question == "61e. Please state your level of agreement with the following statement: Access rights to the managed access area have been distributed fairly to fishers") {
         
         hhs_Q61f <- selectedData()[,c("ma_name", "61f_rights_distribution_fair")] %>%
                           filter(`61f_rights_distribution_fair` %in% c(1:5)) %>%
                              rbind(c(NA,1),c(NA,2),c(NA,3),c(NA,4),c(NA,5))

For Micronesia #14 and #44 a & b are showing Kitti results only.

@Court78 I'm working on #12 but each question probably needs its own issue. I'm starting with 15.

In 15_activity there are 1678 different activities. In many cases "different" activities are just spelling or punctuation differences. Below are a few examples.

You requested that I use "multi answer, bar, mean" but I'm not sure what this means. In looking at the data I wonder if the following would be best:

1. Make everything lower case
2. Identify the top 5 activities
3. Create a faceted bar plot where the facets are the 5 activities

Below I'm showing the top 10 (lower case) activities to give you a sense.

 [987] "memasak"                                                                                   
 [988] "Memasak"                                                                                   
 [989] "MEMASAK"                                                                                   
 [990] "memasak dan mencuci"                                                                       
 [991] "memasak ikan"                                                                              
 [992] "memasak keluarga"                                                                          
 [993] "memasak untuk karyawan koperasi"                                                           
 [994] "memasak untuk keluara"                                                                     
 [995] "memasak untuk keluarga"                                                                    
 [996] "Memasak untuk keluarga"

 1 memasak                      157
 2 keeping the house            138
 3 cooking for family           128
 4 berkebun                     127
 5 farming                      126
 6 agricultura                  116
 7 menjual ikan                 116
 8 merawat anak                 106
 9 caring for children           91
10 labor                         87

@Court78 in the queries for 7_age you should remove invalid ages. I'm handling this in the app right now so no hurry. There are values of 1975, 2001 etc, I assume these are birth years.

@abelvaldivia I'm creating a spec for the input HHS data and am moving away from factors, partly as a way to drop all the as.numeric(as.character calls.

For variable 70_hh_average_income I'm seeing that it's treated as a factor by data.table but that nearly all the values are numbers. And in the app you call as.numeric(as.character.

There are values in one file (HND) like "2500-5000 L". I'm not seeing these handled any special way so it seems that about half of HND values are become NA. Is this the way it should be?

@abelvaldivia there are two choices for "Type of Analysis" but nothing happens if you change it. Is there a plan to do something different?

@abelvaldivia I'm seeing a few plots from the old app that are not in the hhs questions file. Like q45 about leadership positions and two numbered 44 about meetings. Should these be dropped?

Repeats in answer choices. For instance, 3-4 per week and 3-4 times per week.

This function could be cleaned up and made simpler and more flexible.

@abelvaldivia I see that you're removing duplicates based on submission ID and date. Does that mean submissionid can be repeated?

> dim(hhs)
[1] 17286   243
> length(unique(hhs$submissionid))
[1] 17276

@abelvaldivia in the FMA app I used the geographic renaming below so it's clearer what they are, but I'm wondering if perhaps this is a bad idea and I should avoid for HHS? Opinion?

    country,
    iso3 = country_code,
    subnational = level1_name,
    subnational_id = level1_id,
    local = level2_name,
    local_id = level2_id,
    maa = ma_name,
    maa_id = ma_id,

ggplot2

ggplot

@Court78 related to the new plot for question 20. There are 7 gear-related questions. I think what I should do is:

1. For each of the MAA/gear types tally the total number of responses
2. Compute the proportion of responses that used that gear
3. Show a faceted plot with each facet being one of the 7 gear types

So an example would be for "gear hand" for Melekeok, I would tally the total number of non-NA responses (so how many 0 or 1) for the denominator. Then tally the number of 1 values and use this as the numerator.

Do I have this correct?

"20a_gear_hand"                    "20b_gear_stationary_net"         
"20c_gear_mobile_net"              "20d_gear_stationary_line"        
"20e_gear_mobile_line"             "20f_gear_explosives"             
"20g_gear_other"  

@Court78 all the plots in HHS are interactive. I think this is useful but makes styling significantly harder due to the way they are made interactive. Plotly takes a ggplot2 plot and tries to make it interactive and there are some things this transition does not do well. For example, the strip labels do not look good and it will take work to make them look good.

If we used static plots we could use the same style as FMA for the plots and it's easier to make them look good.

But I don't know how important it is to users for it to be interactive.

When I re-wrote the app, I made it so I can change interactive to static for all plots at once changing one FALSE to TRUE so I could let you take a look if you wanted.

I'm seeing what appear to be outliers in the old data that do not show up in the new. I assume these are fixes and not issues with the new data. This results in a different looking plot in the new app and I'm going to assume this is correct.

# Old app
dplyr::filter(hhs_Q68, ma_name == "Kaledupa") %>% 
  dplyr::arrange(dplyr::desc(`68_fish_sold`))

# A tibble: 247 x 3
   ma_name  `68_fish_eaten` `68_fish_sold`
   <fct>              <int>          <int>
 1 Kaledupa              10        1000000
 2 Kaledupa              15         500000
 3 Kaledupa               5         200000
 4 Kaledupa              25            500
 5 Kaledupa              10            200
 6 Kaledupa              20             85
 7 Kaledupa              30             80
 8 Kaledupa              45             80
 9 Kaledupa              40             80
10 Kaledupa              25             65

But in the "new" app I'm seeing the same query yield

# New app
# A tibble: 247 x 3
   maa      `68_fish_eaten` `68_fish_sold`
   <chr>              <dbl>          <dbl>
 1 Kaledupa             100            500
 2 Kaledupa              10            200
 3 Kaledupa              10            100
 4 Kaledupa              20             85
 5 Kaledupa              45             80
 6 Kaledupa              40             80
 7 Kaledupa              30             80
 8 Kaledupa              25             65
 9 Kaledupa              25             60
10 Kaledupa              29             60

@zross There are at least 17 HHS questions that are still not in the app. These questions should be included and are:
Q7, Q15, Q20, Q21, Q30e-Q30i, Q35, Q48, Q49, Q50, Q57, Q58, Q62, Q69, Q71, Q74, Q75, Q76. In the hhs_questions file, these questions are under the column "question_no_included". To add the question to the select menu in the UI, just move the question from the "question_no_included" column to the "question" column. For each question, a code should be written to get the summary plot and summary table.

In old app there is a line break manually inserted. This should be replaced with a max title width before wrapping.

@abelvaldivia in number Q59, I think there is an error in the field renaming. Do you agree? Can you give me the crosswalk for names from Q59_summary to the names you want that I should be using?

The Q59_summary looks like this:

                   MA name    N   Certain Confident.not High.chance  Uncertain Very.confident.not
1  Binongko Makoro Taipabu  270         0          87.8         0.7        8.1                3.3
2                 Kaledupa  280       1.4          55.4        25.4       13.9                3.9
3                Kapuntori  277      14.1          58.5         0.7        2.9               23.8
4                 Kulisusu  276         0          77.2         0.4         21                1.4
5                 Labengki  237       1.3          54.4         5.9       34.2                4.2
6                  Maginti  280      19.6          37.5        19.6       20.4                2.9

But then you rename with:

       colnames(Q59_summary) <- c("MA name", "N", 
                                    "Not very confident",
                                    "Not confident",
                                    "Uncertain",
                                    "Confident", 
                                    "Very confident")

FULL CODE

         hhs_Q59 <- selectedData()[,c("ma_name","59_food_procurement")] %>%
                        filter(`59_food_procurement` != "") %>%
                           rbind(c(NA, "Confident not"), 
                                 c(NA, "Uncertain"),
                                 c(NA, "High chance"),
                                 c(NA, "Very confident not"),
                                 c(NA, "not Certain")) %>%
                           droplevels()
        
         Q59_summary <- proportion (hhs_Q59$`59_food_procurement`,
                                     hhs_Q59$ma_name,
                                     3,5)
         colnames(Q59_summary) <- c("MA name", "N", 
                                    "Not very confident",
                                    "Not confident",
                                    "Uncertain",
                                    "Confident", 
                                    "Very confident")

@zross The following questions are either not plotting or the plot has a format issue

24,39,41,51b,51c,51d,51e,70a,70b

@abelvaldivia an example would be 66. Is it OK if I add an "a" and "b" to distinguish those. I think there are one or two other examples.

@abelvaldivia can you explain why Q14, Q15, Q44 etc are their own datasets rather than being additional columns in the ALL_hhs? It's intensive having to filter each of these when the user makes changes to geography and it's a bit of a pain to use multiple datasets.

@Court78 personally I find it a bit cumbersome to try and guess what category a question will fall under and often have to hunt and peck.

A possible alternative would be a single selector with a text divider. In the image below you can see that this is partly implemented already, a single selector would look like the "Question" selector but would have all questions.

But I'm not sure if this would be worse or better for you and your users. To get to question, say, 70, you would need to scroll through them all which is a separate possible annoyance.

It would take an hour or two to implement.

@abelvaldivia we can skip this tab for HHS right?

@abelvaldivia I think it's confusing in the current HHS app that the selector is separate (but linked) to the radio buttons. I think a collapsible selector might make more sense.

Closed

Open

When I select Q40 it makes me reload the app...

51a, 51b, 51c as well as 51d and 51e

@zross The app is not working correctly on the server, even though it works fine when I run it locally. This start happening after I updated my Mac to the new OS (BigSur). I haven't figured out where the issue is. I am updating the OS again, it looks like Apple just release another OS update, but I am not even sure this is the root of the problem

Add option to show mean and standard error across MA areas selected

@abelvaldivia I'm changing my mind about the collapses and I think an approach like this is better for a couple of reasons:

1. It's confusing that the household data summary is not really a question and all the others show charts.
2. With the collapses it might become confusing what is selected if you close the collapse you're working with
3. I think a user might want to jump between the data summary and a chart.

The design here is not great, but what do you think of this instead? A separate tab for the data and then for the charts. There are no selects on the data tab.

@Court78 I'm not sure if having the y-axis start with the last maa alphabetically was deliberate. I'm guessing no. I reversed it. If this is OK please close this.

@abelvaldivia in q66 the Kulisusu MA has all NA values so I think this MA should be excluded from the plot since "Proportion of community members" is not meaningful if no values are 0 or 1. But in the original app this is shown as a 0.

How should it be handled?

Maybe for geo selectors but definitely for question selector