dev/affwerx_sam.R

In abresler/govtrackR: Federal government tracking tools

library(tidyverse)
library(govtrackR)
library(asbmisc)

setwd("~")
df_affwerx <-
  read_csv("Desktop/deals/sheldon/affwerx/data/affwerx_companies.csv")

df_completed <-
  "Desktop/r_packages/govtrackRdata/data/affwerx/sam_duns/" %>% read_rda_files() %>%
  select(-matches("dataDFAR", "dataFAR"))

df_completed %>%
  select(-matches("data")) %>%
  View()

urls <- df_completed %>% filter(!is.na(urlCompany)) %>% pull(urlCompany)

df_crux <-
   crux_urls(urls = urls)

df_crux

df_urls <-
  df_crux %>%
  select(urlCompany = url, urlImage = image_url, urlIcon = favicon_url) %>%
  mutate(urlLogo = urlImage,
         urlLogo = ifelse(is.na(urlLogo), urlIcon, urlLogo),
         urlLogo = ifelse(is.na(urlLogo), "http://icons.iconarchive.com/icons/icons8/windows-8/512/Programming-Help-icon.png", urlLogo)) %>%
  select(urlCompany, urlLogo)

df_affwerx <-
  df_affwerx %>%
  left_join(df_urls)


# plot --------------------------------------------------------------------
df_plot <-
  df_affwerx %>%
  left_join(df_completed %>% select(-urlCompany), by = "idDUNS")

df_tip <- tribble(
  ~ label,
  ~ column_name,
  ~ url,
  "Company",
  "nameCompany",
  "urlCompany",
  "DUNS",
  "idDUNS",
  NA,
  "Date Founded",
  "dateBusinessStart",
  NA,
  "Date SAM Registration",
  "datetimeRegistration",
  NA,
  "Days Post Founding Until Government Registration",
  "daysPostFoundingPriorToGovernmentWork",
  NA
)

df_plot <-
  df_plot %>%
  mutate(urlLogo = ifelse(is.na(urlLogo), "http://icons.iconarchive.com/icons/icons8/windows-8/512/Programming-Help-icon.png", urlLogo))

df_plot <- table_tooltip(
  data = df_plot,
  thumbnail_parameters = list(
    thumbnail_url = "urlLogo",
    alt =
      "nameCompany",
    height = 80,
    width = 80,
    is_sizing_px = T,
    html_parent =
      "td"
  ),
  df_tip = df_tip,
  span_style = "font-size:10px font-family:lucida;"
)


df_plot <-
  df_plot %>%
  filter(!is.na(idDUNS)) %>%
  filter(!is.na(daysPostFoundingPriorToGovernmentWork)) %>%
  select(idDUNS, nameCompany, datetimeRegistration, daysPostFoundingPriorToGovernmentWork, htmlTooltip)


df_plot %>%
  mutate(
    dateRegistered = as.Date(datetimeRegistration),
    countDaysPostFoundingPriorToGovernmentWork = daysPostFoundingPriorToGovernmentWork
  ) %>%
  hc_xy(
    x = "dateRegistered",
    y = "countDaysPostFoundingPriorToGovernmentWork",
    theme_name = "tufte",
    link = "urlCompany",
    title = "SAM Registered AFWERX Companies",
    tooltip = "htmlTooltip",
    override_y_label = "Log10 Days Post Founding from SAM Registration",
    override_x_label = "Date SAM Registration",
    export = T,
    disable_legend = T,
    enable_point_select = T,
    point_size = 2.5,
    subtitle = "Date of Government Registration by Length of Time in Operation Until Government Registration",
    credits = "195 of 221 companies",
    use_navigator = T,
    file_path = "Desktop/abresler.github.io/asbc_projects/shelden/affwerx/visualizations/days_until_sam_registration",
    transformations = c("log_y", "loess")
  )



# full data ---------------------------------------------------------------


all_data <-
  df_affwerx %>%
  left_join(df_completed %>%
  nest(-idDUNS, .key = "dataSAM") %>%
  mutate(hasSAMData = T)) %>%
  mutate(hasSAMData = ifelse(is.na(hasSAMData), F, hasSAMData)) %>%
  select(idDUNS, nameCompany, hasSAMData, urlCompany, dataSAM) %>%
  arrange(desc(hasSAMData))

all_data


# naics_matches -----------------------------------------------------------

df_naics <-
  df_completed %>%
  filter(hasNAICS) %>%
  select(idDUNS, nameCompanyLegal, dataNAICS) %>%
  mutate_if(is.character, str_trim) %>%
  unnest()

df_opportunities <- fbo_opportunities()

df_contracts <- df_opportunities %>% sample_n(1500)
df_sample_contracts <- parse_fbo_solicitation_urls(df_contracts$urlSolicitationDetails)