metropolitan-council / travel-study-stories Goto Github PK

"Showing data for [x] households based on your selection of [hhs restricted to MPO region], [hh county], [hh city/township]. Use caution when interpreting data with small sample sizes."

Group together all under 18

Two way table should be different variables

• Trip purpose
• Person race
• Households w/ POC
• etc.

currently, factor variables give NA for mean/median
could add an output that is like "mode" or some such
tab_mean <-
data.frame(
mean = NA,
mean_se = NA,
median = NA,
median_se = NA
)

Using waiter - https://github.com/JohnCoene/waiter

Gray out cells in table with

less than 30 respondents for a cell
less than 60 respondents for a table

We need to come up with a list of 5-8 tasks for user volunteers to complete in User Zoom sessions.

One-way:

• An estimated ___
• if percent: "% of"
• [trips/people/households/vehicles]
• are
• value_label
• .

Two-way:

• trip -> hh: __% of [trip value_label] trips are made by [hh value_label] households.

• trip -> per: __% of [trip value_label] trips are made by [per value_label] people.

• trip -> veh: __% of [trip value_label] trips are made with [veh value_label] vehicles.

• trip -> trip: __% of [trip 1 value_label] trips are also [trip 2 value_label]

• trip -> day: __% of [trip value_label] trips are made by people who [day value_label]

• veh -> hh: __% of [veh value_label] vehicles are owned by [hh value_label] households.

• veh -> per: __% of [veh value_label] vehicles are owned by [per value_label] people.

• veh -> veh: __% of [veh 1 value_label] vehicles are also [veh value_label].

• veh -> trip: __% of [veh value_label] vehicles are used on [trip 2 value_label] trips.

• veh -> day: __% of [veh value_label] vehicles are used by people who [day value_label].

• per -> hh: __% of [per value_label] people reside in [hh value_label] households.

• per -> per: __% of [per 1 value_label] people are also [per 2 value_label].

• per -> veh: __% of [per value_label] people own [veh value_label] vehicles.

• per -> trip: __% of [per value_label] people make [trip value_label] trips.

• per -> day: __% of [per value_label] people are also those that [day value_label].

• hh -> hh: __% of [per value_label] people reside in [hh value_label] households.

• hh -> per: __% of [per 1 value_label] people are also [per 2 value_label].

• hh -> veh: __% of [per value_label] people own [veh value_label] vehicles.

• hh -> trip: __% of [per value_label] people make [trip value_label] trips.

• hh -> day: __% of [per value_label] households include people who [day value_label].

in one way table, would want a total sample size at the bottom
in two-way table, want row & column totals

One-way table:

• Estimated
• [percent/number]
• of
• [households/people/trips/vehicles]
• in each
• [variable]
• category.

Two-way table:

• Estimated
• [percent/number]
• of
• [column weighting var: households/people/trips/vehicles]
• in each
• [crosstab_column]
• category, shown separately for each
• [crosstab_row].

wouldn't it be great

could additionally add download button to export the entire dataset filtered to the county/city from filters (maybe under the go_one_way button)

• Filter trips that live in ___ (city), and trips that end in ___ (city)
• Counties, big municipalities

In PSRC data explorer, looks like this:

The data is formatted so that each race is an independent yes/no variable. Can we consolidate this like we have for other demographics?

maybe I need to turn off "ignoreNULL" in this part of mod_filters_oneway.R?

    observeEvent(input$oneway_input_counties,
                 {
                   if (!is.null(input$oneway_input_counties))
                   {
                     filtered_cities <- tbi_tables$hh %>%
                       dplyr::filter(hh_county %in% input$oneway_input_counties) %>%
                       dplyr::select(hh_city) %>%
                       unique() %>%
                       na.omit()

                     updateSelectInput(
                       session = session,
                       inputId = "oneway_input_cities",
                       label = "Household City/Township",
                       choices = filtered_cities$hh_city,
                       selected = NULL
                     )
                   } else {
                     selectInput(
                       session = session,
                       inputId = ns("oneway_input_cities"),
                       "Household City/Township",
                       choices = unique(na.omit(tbi_tables$hh$hh_city)),
                       selected = NULL)
                   }},
                 ignoreInit = TRUE
    )

Group

• Format table percentages, visual options
• Format plot axes, titles
• Pretty Table headers

Originally posted by @ashleyasmus in #38 (comment)

in mod_input_category_col.R:

At line 61, I've turned on the print statement to see what category & variable the app thinks is being selected:

    # print the selected input$category and input$variable
    observe({
      print(paste0(id, "-", input$category, ":", input$variable))
    })

Even though I have initialized the selectInput to reference the variable purpose_category, it's finding bike_park_loc. I'm not sure why.

mod_input_category_col_ui <- function(id){
  ns <- NS(id)
  tagList(
    selectInput(
      inputId = ns("category"),
      label = "Category",
      choices = unique(tbi_dict$category),
      selected = "Trips"
    ),
    selectInput(
      inputId = ns("variable"),
      label = "Variable",
      choices = input_list$Trips,
      selected = "Trip purpose - broad categories"
    )

    # textOutput(outputId = "question")
  )
}

Checking the input_list.rda, It really seems like it should be finding the right variable:

load("~/MetC_Locals/MTS/travel-study-stories/travel.survey.explorer/data/input_list.rda")
input_list$Trips
input_list$Trips["Trip purpose - broad categories"]

maybe above the plot?

chi square analysis running in background?
"Chi square indicates NSD"
Maybe all bars are same color of GRAY if no sig diff among categories

Low-priority issue, but it would be nice if we could remove/replace non-ASCII characters

See potential solution https://stackoverflow.com/a/37254905

step by step, how to use this tool and what it means

Demographics in rows
everything else in column

see, e.g., day table for reasons for not traveling that day

@JonathanEhrlichMC suggested a radio button to indicate 2019 or 2021 survey