Build Adobe Analytics Segments Using R - adobeanalyticsr v0.3.0

Define the Rule(s)

Rules are made up of 3 main elements: subject (dimension/metric), verb, and object.

Rule

DT::datatable(seg_verbs, 
              extensions = c('FixedColumns',"FixedHeader"),
              options = list(scrollX = F, 
                             paging=T,
                             fixedHeader=T))

rule1 <- seg_rule(dimension = 'page', 
                  verb = 'streq',
                  object = 'Home')

## {
##   "func": "streq",
##   "str": "Home",
##   "val": {
##     "func": "attr",
##     "name": "variables/page"
##   }
## }

rule2 <- seg_rule(dimension = 'daterangeday',
                  verb = 'eq',
                  object = '2022-01-25')

rule3 <- seg_rule(dimension = 'mobiledevicetype',
                  verb = 'eq',
                  object = 2163986270)

Create the Container

Every segment has at least one container. Adding an additional container to the main container is only needed if you are using a different “conjunction” between then. Here I am creating a “visit” context container using “rule1” and “rule2” and using the conjunction “and”. This will result in a segment that includes visits where both rules are true.

container1 <- seg_con(context = 'visits',
                      conjunction = 'and',
                      rules = list(rule1, rule2))

Build the Segment

Now that we have the rules and containers defined we need to build the segment using the seg_build() function. By default, this will return a JSON string that can then be validated using the seg_val() function but you are able to change the argument “create_seg” to TRUE and it will create the segment in the UI and return the segment id which can then be used in your other API calls.

In the following example the segment name is “Home Page visits” and the description is “All visits that include a page named ‘Home’”. I am using the “containers” argument because we have at least one container to be included in our segment. If we were just using rules then we would use the “rules” argument.

built <- seg_build(name = "Home Page visits",
                   description = "All visits that include a page named 'Home'",
                   containers = list(container1, rule3),
                   context = 'visits',
                   conjunction = "or",
                   create_seg = FALSE)

Since the argument create_seg() is set to FALSE, the response of the function is a json string. As you can see, segment definitions can become quite complex to debug but going through each step of the process can help you quickly identify any issues or areas you can improve on. But now you also have a clear understanding of the definition and can easily include this as part of your analysis documentation. This analysis artifact can then be used by your future self or the next analyst.

## {
##   "name": "Home Page visits",
##   "description": "All visits that include a page named 'Home'",
##   "definition": {
##     "func": "segment",
##     "version": [1, 0, 0],
##     "container": {
##       "func": "container",
##       "context": "visits",
##       "pred": {
##         "func": "or",
##         "preds": [
##           {
##             "func": "container",
##             "context": "visits",
##             "pred": {
##               "func": "and",
##               "preds": [
##                 {
##                   "func": "streq",
##                   "str": "Home",
##                   "val": {
##                     "func": "attr",
##                     "name": "variables/page"
##                   }
##                 },
##                 {
##                   "func": "eq",
##                   "val": {
##                     "func": "attr",
##                     "name": "variables/daterangeday"
##                   },
##                   "num": 1220025
##                 }
##               ]
##             },
##             "val": {}
##           },
##           {
##             "func": "eq",
##             "pred": {
##               "preds": {}
##             },
##             "num": 2163986270,
##             "val": {
##               "func": "attr",
##               "name": "variables/mobiledevicetype"
##             }
##           }
##         ]
##       }
##     }
##   },
##   "rsid": "ageo1xxpnwsdi2020prod"
## }

Finally, let’s make sure this segment will validate in the report suite. It is important to note that the validation function uses the segment validation endpoint which doesn’t confirm the segment definition is returning any results, it simply validates that the segment elements are able to be used in the way you have compiled them. For instance, if you were to use “mobiledevice” instead of “mobiledevicetype” then the result of seg_val(built) would return an error stating the dimension was not able to be found.

If the segment validates it should look like this:

seg_val(built)

## [1] "The segment is valid."

Conclusion

We have already been using this in our analysis and data science workflows and it has proven to be a big help. In one scenario, I had to create 13 different complex segments based on the defined segments we wanted to isolate in our analysis. After creating them all in R I presented the initial findings to the team who quickly noticed I was missing an exclusion rule that should have been included on all 13 different segments. Normally the process would have been a huge headache but thankfully, I had coded all the segments and adding one simple exclusion rule to each one was quick. In addition, I was able to move forward with more confidence in my segments because using code eliminates much of the errors that happend due to multiple touch points in generating an analysis.