Answers questions like how many sites have certain indicators >2x the state avg?

This function provides tables of summary stats but also text that explains those findings in plain English. It relies on colcounter_summary_all()

Usage

count_sites_with_n_high_scores(
  scores,
  thresholds = NULL,
  indicator_type = c("percentile", "ratio", "other")[2],
  xwide = c("", "statewide", "nationwide")[1],
  site_stat_type = c("cum_pct", "cum_count", "pct", "count")[1],
  text_suffix = c("th percentile", " times the average", "")[match(indicator_type,
    c("percentile", "ratio", "other"))],
  text_indicatortype = "indicators",
  quiet = !interactive()
)

Arguments

scores

scores in a table with one row per place and one column per indicator

thresholds

thresholds vector of numbers as benchmarks. Assuming the indicators in the scores table are ratios to the average, then the thresholds could be for example, 1.50, 2, etc. which would represent ratios that are 1.5x or 2x etc.

indicator_type

Scores and benchmarks are percentiles, ratios, or anything else. One of these: c("percentile", "ratio", "other")

xwide

must be "statewide" or "nationwide" – used only in the text output that describes the findings.

site_stat_type

Count or share of sites, exactly or cumulatively. One of these: c("cum_pct", "cum_count", "pct", "count")

text_suffix

If using ratios, use the default, which explains these thresholds as X times the average. If using percentiles as thresholds, set text_suffix = "." or text_suffix = "th percentile in the state." for example

text_indicatortype

can be "Summary Indexes" etc.

quiet

whether to print findings to console

Value

Returns a list with two named elements, "stats" and "text" where stats is a 3-dimensional array of numbers. See dimnames(output$stats).

Details

Helps provide stat summaries such as:

(x%) of these (sites) have

at least (N) of these (YTYPE )indicators

at least (R) times the (State/National average)

Examples

# out <- ejamit(testpoints_100, radius = 1)
out <- testoutput_ejamit_1000pts_1miles 
x <- out$results_bysite
x <- setDF(copy(x))
#> Error in setDF(copy(x)): could not find function "setDF"
ratio_data <- x[, names_d_ratio_to_state_avg]
#> Error in `[.data.table`(x, , names_d_ratio_to_state_avg): j (the 2nd argument inside [...]) is a single symbol but column name 'names_d_ratio_to_state_avg' is not found. If you intended to select columns using a variable in calling scope, please try DT[, ..names_d_ratio_to_state_avg]. The .. prefix conveys one-level-up similar to a file system path.
ratio_benchmarks <- c(2, 3, 5, 10)

findings <- count_sites_with_n_high_scores(ratio_data, ratio_benchmarks)
#> Error: object 'ratio_data' not found

names(findings)
#> Error: object 'findings' not found
dim(findings$text)
#> Error: object 'findings' not found
#   see most striking stat only
tail(findings$text[findings$text != ""], 1) # the most extreme finding
#> Error: object 'findings' not found
findings$text[findings$text != ""]
#> Error: object 'findings' not found

dimnames(findings$stats) # count, cut, stat
#> Error: object 'findings' not found
## stat can be count, cum, pct, or cum_pct

findings$stats[1,,] # any of the indicators (at least one indicator)
#> Error: object 'findings' not found
findings$stats[,,"count"]
#> Error: object 'findings' not found
findings$stats[ , , 1]
#> Error: object 'findings' not found
findings$stats[ , 1, ]
#> Error: object 'findings' not found

pctile_data <- testoutput_ejamit_1000pts_1miles$results_bysite
pctile_data <- pctile_data[, ..names_ej_state_pctile]
pctile_benchmarks <- 90
y <- count_sites_with_n_high_scores(pctile_data, pctile_benchmarks,
  indicator_type = "percentile",
  text_indicatortype = "Summary Indexes", xwide = "statewide" 
  )

# At how many sites is at least one of these indicators at least 90th pctile nationwide?
sum(rowMaxs2(data.frame(pctile_data)) >= 90, na.rm = T)
#> Error in rowMaxs2(data.frame(pctile_data)): could not find function "rowMaxs2"
## or 
count_sites_with_n_high_scores(pctile_data, 90, xwide = "nationwide", quiet = T)$stats[count = "1", cut = "90", stat = "cum"] 
#> [1] 268

# see most striking stat only
mx <- count_sites_with_n_high_scores(pctile_data, 
  thresholds = 1:100, quiet = TRUE,
  text_indicatortype = "Summary Indexes",
  text_suffix = "th percentile in the state.")
tail(mx$text[mx$text != ""], 1) # the most extreme finding
#> [1] "At at least 1% of these sites, 2 of the Summary Indexes are 99th percentile in the state. "