31:T56b,```r meta <- readRDS(url("https://rud.is/data/tags.rds")) tags <- meta$metadata table( tags$created_at, dnn = "created_at" ) |> as.data.frame.table( responseName = "n", stringsAsFactors = FALSE ) |> transform( created_at = as.Date(created_at) ) -> xdf data.frame( created_at = seq.Date(min(xdf$created_at), max(xdf$created_at), "1 day") ) |> merge( xdf, all.x = TRUE ) |> transform( n = ifelse(is.na(n), 0, n), days_elapsed = as.numeric(created_at - min(created_at)) ) |> transform( csum = cumsum(n) ) -> xdf poisson_model <- glm(csum ~ days_elapsed, data = xdf, family = poisson(link = "log")) target_csum <- 2000 predicted_days_elapsed <-xdf$days_elapsed predictions <- vector("list", 2000) p_idx <- 0 while(TRUE) { predicted_days_elapsed <- max(predicted_days_elapsed) + 1 predicted_csum <- predict(poisson_model, newdata = data.frame(days_elapsed = predicted_days_elapsed), type = "response") p_idx <- p_idx + 1 data.frame( created_at = min(xdf$created_at) + predicted_days_elapsed, predicted_count = as.integer(predicted_csum) ) -> predictions[[p_idx]] if (predicted_csum >= target_csum) break } predicted_date <- min(xdf$created_at) + as.integer(predicted_days_elapsed) ydf <- do.call(rbind.data.frame, predictions) ydf$created_at <- as.character(ydf$created_at) ydf ```32:[] 2f:{"title":"WebR glm() In Observable","cells":[{"id":0,"value":"# WebR `glm()` In Observable","pinned":false,"hidden":false,"mode":"md"},{"id":112,"value":"I keep tabs on [our tag (detection rule) inventory](https://viz.greynoise.io/tags), and one of those tabs is a simple `glm()` model that helps me estimate when we'll be at `n` tags.\n\nI used to do that in a `cron` job, but now I can do that in Observable with R vs. convert to some janky JS stats workflow.\n\n[Scroll down](https://observablehq.com/d/b8336b39e60d2ca0#rPredict) for the R code/explanation.","pinned":false,"hidden":false,"mode":"md"},{"id":20,"value":"html`⏳ WebR is loading…`","pinned":false,"hidden":false,"mode":"ojs"},{"id":101,"value":"Plot.plot({\n x: {\n grid: true\n },\n y: {\n grid: true\n },\n marks: [\n Plot.lineY(predicted, { x: \"created_at\", y: \"predicted_count\", tip: true })\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":89,"value":"predicted","pinned":true,"hidden":false,"mode":"ojs"},{"id":83,"value":"predicted = webRDataFrameToJS(await rPredict.toJs()).map((d) => ({\n created_at: new Date(d.created_at),\n predicted_count: d.predicted_count\n}))","pinned":true,"hidden":false,"mode":"ojs"},{"id":39,"value":"rPredict = await webR.evalR(predictR.innerText)","pinned":true,"hidden":false,"mode":"ojs"},{"id":129,"value":"We can put the R code into a Markdown code block so we get syntax highlighting and somewhat less-janky editing.\n\nThe cell above this one just grabs the inner text of the block to do the eval.\n\nI'm just using base R idioms to avoid layering on package dependencies.","pinned":false,"hidden":false,"mode":"md"},{"id":116,"value":"$31","pinned":false,"hidden":false,"mode":"md"},{"id":56,"value":"import { webRDataFrameToJS } from \"@hrbrmstr/webr-0-2-2-in-observable\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":15,"value":"webR = {\n const { WebR, ChannelType } = await import(\n \"https://cdn.jsdelivr.net/npm/webr/dist/webr.mjs\"\n );\n\n const intWebR = new WebR({\n channelType: ChannelType.PostMessage,\n interactive: false\n });\n\n await intWebR.init();\n\n document.querySelector(\"#loader\").innerText = \"🟢\";\n document.querySelector(\"#msg\").innerText = \"WebR loaded!\";\n\n return intWebR;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":3,"value":"import { style } from \"@hrbrmstr/hrbrthemes\"","pinned":false,"hidden":false,"mode":"ojs"},{"id":11,"value":"html`${style}`","pinned":false,"hidden":false,"mode":"ojs"}],"files":"$W32","stdlib":"1","titling":"auto"}