aggregating.rmd

---
title: Aggregating event data
output: 
  html_document:
    toc: true
    toc_float: 
      collapsed: false
      print: false
---

```{r include = F}
library(bupaR)
```

```{r echo = F}
htmltools::includeHTML("tracking_google_analytics.html")
```

`bupaR` supports several strategies to zoom-out from too detailed activities. One option is to remove distinctions between similar activity types; changing them to a uniform name. A second option is to collapse activities that belong together as a sub process under a higher-level name. The first aggregation we call a "IS-A" aggregation and the second a "PART-OF" aggregation. 

## Is-a aggregation

As a simple example, consider the `patients` event data.

```{r fig.width = 8}
patients %>% 
	process_map()
```

In the hypothetical situation where we want to zoom out this process, we could say that both MRI SCAN and X-Ray are _Scans_. That is, a MRI SCAN __is a__ Scan and a X-Ray __is a__ Scan. As a result, we could perform a is-a aggregation. For this, we use the `act_unite` function, since we `unite` two or more activities. We will see that the 236 MRI Scans and 261 X-Rays are replaced with 497 Scans. 

```{r fig.width = 8}
patients %>%
	act_unite(Scan = c("MRI SCAN","X-Ray")) %>%
	process_map()
```

## Part-of aggregation

Another approach is to combine activities which are not simular, but belong together as _part of_ a sub process. Suppose we would say that X-Ray, MRI Scan, and Blood test are _part of_ the sub process "Testing". We could _collapse_ the occurence of these activities into a single activity. This can be achieved with the `act_collapse` function.

```{r fig.width = 8}
patients %>% 
	act_collapse(Testing = c("MRI SCAN","X-Ray","Blood test")) %>%
	process_map()
```



## Recoding individual activities

Sometimes it is useful to recode individual activity levels, for instance when some typo's are present, or when you want to give different labels a more uniform format. Individual recodings can be done using `act_recode`.

```{r fig.width = 8}
patients %>%
	act_recode("Check-in" = "Registration",
			   "MRI Scan" = "MRI SCAN") %>%
	process_map()
```