01-startup.Rmd

```{r}
###################################################################################################
##
## STARTUP CODE TO LOAD LIBRARIES AND DATA
## 
## NO TEXT
##
###################################################################################################
```


```{r 01-su-clean-load,echo=FALSE,child="CleanLoad.Rmd"}
```


<!-- 
```{r 01-su-needed-functions}
############################################################################################################
##
## Given a date and a 3-digit zip, returns a triplet of:
##  - the unemployment rate
##  - the median income
##  - personal income
## for that ZIP.
##
## The indicators are published by FIPS (Federal Information Processing Standard state code). A ZIP code
## can cover several FIPS. In such a case, the function looks up all the relevant FIPS code and returns
## a population weighted average of the indicators.
##
##

indicatorsFromZIPandDate <- function(zip3d, application_date) {
  require(tidyverse)
  require(lubridate)
  
  emptyResult <-
    tibble(
      unemployment = 0,
      medianIncome = 0,
      personalIncome = 0
    )
  
  
  if (!is.numeric(zip3d))
    return(emptyResult)
  if (!is.Date(application_date))
    return(emptyResult)
  
  # Get all the FIPS with population
  populatedzips <- zipfips %>%
    # Look up a specific zip
    filter(zip == zip3d) %>%
    select(zip, population, county_fips) %>%
    
    # Rename the FIPS column
    rename(FIPS = county_fips) %>%
    
    # Add the population of all the counties with the same FIPS
    group_by(FIPS) %>%
    mutate(population = sum(population)) %>%
    
    # And keep unique zip/FIPS pairs with its population
    distinct(FIPS, .keep_all = TRUE)
  
  
  # Get macro stats
  ratePerFIPS <- unemploymentRate %>%
    # filter for FIPS
    filter(FIPS %in% populatedzips$FIPS) %>%
    
    # And all dates before the application date
    filter(application_date >= Date) %>%
    
    # Sort the dates in decreasing order
    arrange(desc(Date)) %>%
    
    # So that we only keep the first one
    distinct(FIPS, .keep_all = TRUE)
  
  medianPerFIPS <- medianIncome %>%
    filter(FIPS %in% populatedzips$FIPS) %>%
    filter(application_date >= Date) %>%
    arrange(desc(Date)) %>%
    distinct(FIPS, .keep_all = TRUE)
  
  personalPerFIPS <- personalIncome %>%
    filter(FIPS %in% populatedzips$FIPS) %>%
    filter(application_date >= Date) %>%
    arrange(desc(Date)) %>%
    distinct(FIPS, .keep_all = TRUE)
  
  # Put it all back together
  summ <- populatedzips %>%
    left_join(ratePerFIPS, by = "FIPS") %>%
    left_join(medianPerFIPS, by = "FIPS") %>%
    left_join(personalPerFIPS, by = "FIPS") %>%
    mutate_at(c("unemploymentRate", "medianIncome", "personalIncome"),
              as.double)
  
  # And return
  tibble(
    unemployment = sum(summ$population * summ$unemploymentRate) / (sum(summ$population)),
    medianIncome = sum(summ$population * summ$medianIncome) / (sum(summ$population)),
    personalIncome = sum(summ$population * summ$personalIncome) / (sum(summ$population))
  )
}

```
-->