03-Calculating-SIMD.Rmd

# Calculating SIMD {#simd}

For the final calculation of the SIMD ranks we follow a few simple steps, see the script `scripts/calculations/openSIMD.R`:

- load packages and data
- inverse domain rankings
- exponentially transform the inversed domain rankings
- combine transformed domain scores
- rank final SIMD scores

First, we load the packages and data from the previous domain rank calculations.

```{r}
library(dplyr)
source("../openSIMD_analysis/scripts/utils/helpers.R")
domains <- read.csv("../openSIMD_analysis/results/domain_ranks.csv")
```

Then we invert the domain rankings (so that 1 = least deprived) and exponentially transform them, using [`expoTransform`](#expoTransform). This step spreads out the scores at the deprived end of each domain, so that where an area is deprived in one domain, this deprivation won't be cancelled out by a lack of deprivation in another domain.

```{r}
invRank <- function(v) rank(-v)

exponential_domains <- domains %>%
  mutate_at(vars(-data_zone), funs(invRank)) %>%
  mutate_at(vars(-data_zone), funs(expoTransform))
```

We then combine the domain ranks using fixed weightings.

```{r}
with(exponential_domains, {
  simd_score <<-
    .28 * income +
    .28 * employment +
    .14 * health +
    .14 * education +
    .09 * access +
    .05 * crime +
    .02 * housing
})
```

Finally, we rank the final SIMD score to get the SIMD rankings (1 = most deprived).

```{r}
simd_rank <- rank(-simd_score)
```

SIMD and domain ranks are saved in `results/domain_ranks.csv` and `results/openSIMD_ranks.csv`.