Creating logs

Transforming your raw data into an event log object is one of the most challenging tasks in process analysis. On this page, we cover all the possible situations and challenges that you can encounter.

We start with some important terminology:

  • Case: The subject of your process, e.g. a customer, an order, a patient.
  • Activity: A step in your process, e.g. receive order, sent payment, perform MRI SCAN, etc.
  • Activity instance: The execution of a specific step for a specific case.
  • Event: A registration connected to an activity instance, characterized by a single timestamp. E.g. the start of Perform MRI SCAN for Patient X.
  • Resource: A person or machine that is related to the execution of (part of) an activity instance. E.g. the radiologist in charge of our MRI SCAN.
  • Lifecycle status: An indication of the status of an activity instance connect to an event. Typical values are start, complete. Other possible values are schedule, suspend, resume, etc.
  • Trace: A sequence of activities. The activity instances that belong to a case will result to a specific trace when ordered by the time each instance occurred.

Logs: eventlog vs activitylog

bupaR supports two different kinds of log formats, both of which are an extension on R data.frame:

  • eventlog: Event logs are created from data.frame in which each row represents a single event. This means that it has a single timestamp.
  • activitylog: Activity logs are created from data.frame in which each row represents a single activity instances. This means it can has multiple timestamps, stored in different columns.

The data model below shows the difference between these two levels of observations, i.e. activity instances vs events.

The example below shows an excerpt of an event log containing 6 events. It can be seen that each event is linked to a single timestamp. As there can be more events within a single activity instance, each event also needs to be linked to a lifecycle status (here the registration_type). Furthermore, an activity instance identifier (handling_id) is needed to indicated which events belong to the same activity instances.

handling patient employee handling_id registration_type time
Registration 492 r1 492 start 2018-04-28 03:05:15
Registration 492 r1 492 complete 2018-04-28 05:10:22
Triage and Assessment 492 r2 992 start 2018-04-29 13:11:47
Triage and Assessment 492 r2 992 complete 2018-04-30 03:58:13
X-Ray 492 r5 1729 start 2018-04-30 11:13:48
X-Ray 492 r5 1729 complete 2018-04-30 15:14:23
Transactional lifecycle?
An event is an atomic registration related to an activity instance. It thus contains one (and only one) timestamp. Additionally, the event should include a reference to a lifecycle transition. More specifically, multiple events can describe different lifecycle transitions of a single activity instance. For example, one event might record when a surgery is scheduled, another when it is started, yet another when it is completed, etc.
The standard transactional lifecycle.

The table below show the same data as above, but now using the activitylog format. It can be seen that there are now just 3 rows instead of 6, but each row as 2 timestamps, representing 2 events. The lifecycle status represented by those timestamps is now the column names of those variables.

handling patient employee handling_id complete start
Registration 492 r1 492 2018-04-28 05:10:22 2018-04-28 03:05:15
Triage and Assessment 492 r2 992 2018-04-30 03:58:13 2018-04-29 13:11:47
X-Ray 492 r5 1729 2018-04-30 15:14:23 2018-04-30 11:13:48

As these examples show, both formats can often be used for representing the same process data. However, there are some important differences between them:

  • the eventlog format has much more flexibility in terms of lifecycle. There is no limit to the number of events that can occur in a single activity instance. If your data contains lifecycle statuses such as suspend, resume or reassign, they can be recorded multiple times within a single activity instance. In the activitylog format, as each lifecycle gets is own column, it isn’t possible to have two events of the same lifecycle status in a single activity instance.
  • the level of observation in an eventlog is an event. As a result, attribute values can be stored at the event level. In an activitylog, the level of observation is an activity instance. This means that all additional attributes that you have about your process should be at this higher level. For example, an activity instance can only be connected to a single resource in the activitylog format, whereas in an eventlog different events within the same activity instance can have different resources, of different values for any other attribute.
  • because of the limited flexibility, an activitylog is easier to make, and typically closer to the format that your data is already in (see further below on how to construct log objects). As a result of this, there are many situations in which the analysis of an activitylog will be much faster compared to eventlog, where a lot of additional complexity needs to be taken into account.

The right log for the job

Functionalities in bupaR core packages support both formats. 1 As such, the goal of your analysis does not impact the decision. Only the complexity of your data is important to make this decision. The precise format your raw data is in will further define the preparatory steps that are needed. We can distinguish between 3 typical scenarios. The flowchart below helps you on your way.

An activitylog is the best option when each row in your data is an activity instance, or when events belonging to the same activity instance have equal attribute values (e.g. all events are executed by the same resource). When these two criteria do not hold, you can create an eventlog object.

Scenario 1

If each row in your data.frame is already an activity instance, the activitylog format is the best way to go. Consider the data sample below.

patient handling activity_started activity_ended
326 Blood test 2017-11-10 17:17:10 2017-11-11 00:02:54
326 Check-out 2017-11-11 16:19:58 2017-11-11 19:13:08
326 Discuss Results 2017-11-11 12:36:35 2017-11-11 16:19:58
326 MRI SCAN 2017-11-11 03:54:47 2017-11-11 07:02:17
326 Registration 2017-11-09 03:47:37 2017-11-09 07:13:13
326 Triage and Assessment 2017-11-09 15:44:32 2017-11-10 07:17:03

As each row contains multiple timestamps, i.e. activity_started and activity_ended, it is clear that each row represents an activity instance. Turning this dataset in an activitylog requires the following steps:

  1. Timestamp variables should be named in correspondence with the standard Transactional lifecycle.
  2. Timestamp variables should be of type Date or POSIXct.
  3. Use the activitylog constructor function.
data %>%
    # rename timestamp variables appropriately
    dplyr::rename(start = activity_started, 
           complete = activity_ended) %>%
    # convert timestamps to 
    convert_timestamps(columns = c("start", "complete"), format = ymd_hms) %>%
    activitylog(case_id = "patient",
                activity_id = "handling",
                timestamps = c("start", "complete"))
## # Log of 12 events consisting of:
## 1 trace 
## 1 case 
## 6 instances of 6 activities 
## 0 resources 
## Events occurred from 2017-11-09 03:47:37 until 2017-11-11 19:13:08 
##  
## # Variables were mapped as follows:
## Case identifier:     patient 
## Activity identifier:     handling 
## Resource identifier:     employee 
## Timestamps:      start, complete 
## 
## # A tibble: 6 × 5
##   patient handling              start               complete            .order
##   <chr>   <fct>                 <dttm>              <dttm>               <int>
## 1 326     Blood test            2017-11-10 17:17:10 2017-11-11 00:02:54      1
## 2 326     Check-out             2017-11-11 16:19:58 2017-11-11 19:13:08      2
## 3 326     Discuss Results       2017-11-11 12:36:35 2017-11-11 16:19:58      3
## 4 326     MRI SCAN              2017-11-11 03:54:47 2017-11-11 07:02:17      4
## 5 326     Registration          2017-11-09 03:47:37 2017-11-09 07:13:13      5
## 6 326     Triage and Assessment 2017-11-09 15:44:32 2017-11-10 07:17:03      6

Note that in case a resource identifier is available, this information can be added in the activitylog call.

Scenario 2

If each row in your data.frame is an event, but all events that belong to the same activity instance share the same attribute values, the activitylog format is again the best way to go. Consider the data sample below.

patient handling employee handling_id registration_type time
474 Registration r1 474 started 2018-04-11 14:18:54
474 Triage and Assessment r2 974 started 2018-04-11 16:24:04
474 Registration r1 474 completed 2018-04-11 16:24:04
474 Triage and Assessment r2 974 completed 2018-04-12 01:53:27
474 Blood test r3 1228 started 2018-04-12 11:28:30
474 Blood test r3 1228 completed 2018-04-12 16:11:41
474 MRI SCAN r4 1465 started 2018-04-12 20:00:02
474 MRI SCAN r4 1465 completed 2018-04-13 00:40:10
474 Discuss Results r6 2208 started 2018-04-13 05:51:49
474 Discuss Results r6 2208 completed 2018-04-13 08:24:07
474 Check-out r7 2703 started 2018-04-13 15:38:18
474 Check-out r7 2703 completed 2018-04-13 17:23:56

The resource identifier (employee) has been added as an additional attribute. Note that though each row is an event, they can be grouped into activity instances using the handling_id column, which we will call the activity instance id. Using the latter, we can see that the resource attribute is the same within each activity instance, which allows us to create an activitylog. The steps to do so are the following.

  1. Lifecycle variable should be named in correspondence with the standard Transactional lifecycle.
  2. Timestamp variable should be of type Date or POSIXct.
  3. Use the eventlog constructor function.
  4. Convert to activitylog using to_activitylog for reduced memory usage and improved performance.
data %>%
    # recode lifecycle variable appropriately
    dplyr::mutate(registration_type = forcats::fct_recode(registration_type, 
                                                          "start" = "started",
                                                          "complete" = "completed")) %>%
    convert_timestamps(columns = "time", format = ymd_hms) %>%
    eventlog(case_id = "patient",
                activity_id = "handling",
                activity_instance_id = "handling_id",
                lifecycle_id = "registration_type",
                timestamp = "time",
                resource_id = "employee") %>%
    to_activitylog() -> tmp_act

Note that the resource identifier is optional, and can be left out of the eventlog call if such an attribute does not exist in your data. If the activity instance id does not exist, some heuristics are available to generate it: Missing activity instance id.

Scenario 3

If each row is an event, and events of the same activity instance have differing attribute values, the flexibility of eventlog objects is required. Consider the data sample below.

patient handling employee handling_id registration_type time
377 Registration r3 377 started 2017-12-26 19:09:57
377 Registration r1 377 completed 2017-12-26 20:24:26
377 Triage and Assessment r7 877 started 2017-12-27 13:46:05
377 Triage and Assessment r4 877 completed 2017-12-28 05:09:51
377 Blood test r4 1178 started 2017-12-30 11:06:12
377 Blood test r6 1178 completed 2017-12-30 16:14:23
377 MRI SCAN r3 1415 started 2017-12-30 20:24:12
377 MRI SCAN r6 1415 completed 2017-12-31 01:22:41
377 Discuss Results r7 2111 started 2017-12-31 09:30:37
377 Discuss Results r2 2111 completed 2017-12-31 11:46:53
377 Check-out r2 2606 started 2018-01-01 01:48:21
377 Check-out r1 2606 completed 2018-01-01 03:35:35

In this example, different resources (employees) sometimes perform the start and complete event of the same activity instance. Therefore, we resort to the eventlog format which has no problems storing this. The steps to take are the following:

  1. Lifecycle variable should be named in correspondence with the standard Transactional lifecycle.
  2. Timestamp variable should be of type Date or POSIXct.
  3. Use the eventlog constructor function.
data %>%
    # recode lifecycle variable appropriately
    dplyr::mutate(registration_type = forcats::fct_recode(registration_type, 
                                                          "start" = "started",
                                                          "complete" = "completed")) %>%
    convert_timestamps(columns = "time", format = ymd_hms) %>%
    eventlog(case_id = "patient",
                activity_id = "handling",
                activity_instance_id = "handling_id",
                lifecycle_id = "registration_type",
                timestamp = "time",
                resource_id = "employee") 
## Warning in validate_eventlog(eventlog): The following activity instances are
## connected to more than one resource: 1178,1415,2111,2606,377,877
## # Log of 12 events consisting of:
## 1 trace 
## 1 case 
## 6 instances of 6 activities 
## 6 resources 
## Events occurred from 2017-12-26 19:09:57 until 2018-01-01 03:35:35 
##  
## # Variables were mapped as follows:
## Case identifier:     patient 
## Activity identifier:     handling 
## Resource identifier:     employee 
## Activity instance identifier:    handling_id 
## Timestamp:           time 
## Lifecycle transition:        registration_type 
## 
## # A tibble: 12 × 7
##    patient handling   employee handling_id registration_type time               
##    <chr>   <fct>      <fct>    <chr>       <fct>             <dttm>             
##  1 377     Registrat… r3       377         start             2017-12-26 19:09:57
##  2 377     Registrat… r1       377         complete          2017-12-26 20:24:26
##  3 377     Triage an… r7       877         start             2017-12-27 13:46:05
##  4 377     Triage an… r4       877         complete          2017-12-28 05:09:51
##  5 377     Blood test r4       1178        start             2017-12-30 11:06:12
##  6 377     Blood test r6       1178        complete          2017-12-30 16:14:23
##  7 377     MRI SCAN   r3       1415        start             2017-12-30 20:24:12
##  8 377     MRI SCAN   r6       1415        complete          2017-12-31 01:22:41
##  9 377     Discuss R… r7       2111        start             2017-12-31 09:30:37
## 10 377     Discuss R… r2       2111        complete          2017-12-31 11:46:53
## 11 377     Check-out  r2       2606        start             2018-01-01 01:48:21
## 12 377     Check-out  r1       2606        complete          2018-01-01 03:35:35
## # ℹ 1 more variable: .order <int>

Note that we need an eventlog irrespective of which attribute values are differing, i.e. it can be resources, but also any additional variables you have in your data set. For the special case of resource values, it might be that a different resource executing events in the same activity instance is a data quality issue. If so, some functions can help you to identify this issue: Inconsistent Resources.

Again, if the activity instance id does not exist, some heuristics are available to generate it: Missing activity instance id.

Typical problems

Missing activity instance id

In order to be able to correlate events which belong to the same activity instance, an activity instance identifier is required. For example, in the data shown below, it is possible that a patient has gone through different surgeries, each with their own start- and complete event. The activity instance identifier will then allow to distinguish which events belong together and which do not. It is important to note that this instance identifier should be unique, also among different cases and activities.

patient activity timestamp status activity_instance
John Doe check-in 2017-05-10 08:33:26 complete 1
John Doe surgery 2017-05-10 08:53:16 start 2
John Doe surgery 2017-05-10 09:25:19 complete 2
John Doe treatment 2017-05-10 10:01:25 start 3
John Doe treatment 2017-05-10 10:35:18 complete 3
John Doe surgery 2017-05-10 10:41:35 start 4
John Doe surgery 2017-05-10 11:05:56 complete 4
John Doe check-out 2017-05-11 14:52:36 complete 5

If the activity instance identifier is not available you can use the assign_instance_id() function, which uses an heuristic to create the missing identifier. Alternatively, you can try to create the identifier on your own using dplyr::mutate() and other manipulation functions.

Large Datasets and Validation

By default, bupaR validates certain properties of the activity instances that is supplied when creating an event log:

  • a single activity instance identifier must not be connected to multiple cases,
  • a single activity instance identifier must not be connected to multiple activity labels,

However, these checks are not efficient and may lead to considerable performance issues for large data frames. It is possible to deactivate the validation in case you already know that your data fulfills all the requirements, using the argument validate = FALSE when creating the eventlog. Note that when the activity instance id was created with the assign_instance_id() function, you can assume the above properties hold.

Inconsistent Resources

Each event can contain the notion of a resource. It can be so that different events belonging to the same activity instance are executed by different resources, as in the eventlog below.

patient handling employee handling_id registration_type time .order
401 Registration r6 401 start 2018-01-24 19:26:12 1
401 Registration r7 401 complete 2018-01-24 22:02:51 2
401 Triage and Assessment r5 901 start 2018-01-25 04:32:05 3
401 Triage and Assessment r6 901 complete 2018-01-25 13:41:02 4
401 X-Ray r7 1685 start 2018-01-27 14:03:44 5
401 X-Ray r1 1685 complete 2018-01-27 17:10:26 6
401 Discuss Results r2 2135 start 2018-01-29 09:07:13 7
401 Check-out r5 2630 start 2018-01-29 12:05:20 8
401 Discuss Results r2 2135 complete 2018-01-29 12:05:20 9
401 Check-out r1 2630 complete 2018-01-29 13:25:13 10

If you have a large dataset, and want to have an overview of the activity instances that have more than one resource connected to them, you can use the detect_resource_inconsistences() function.

log %>%
    detect_resource_inconsistencies()
## # A tibble: 4 × 5
##   patient handling              handling_id complete start
##   <chr>   <fct>                 <chr>       <chr>    <chr>
## 1 401     Check-out             2630        r1       r5   
## 2 401     Registration          401         r7       r6   
## 3 401     Triage and Assessment 901         r6       r5   
## 4 401     X-Ray                 1685        r1       r7

If you want to remove these inconsistencies, a quick fix is to merge the resource labels together with fix_resource_inconsistencies(). Note that this is not needed for eventlog, but it is for activitylog. While the creation of the eventlog will emit a warning when resource inconsistencies exist, this should mostly be seen as a data quality warning. That said, there might be analysis related to the counting of resources where such inconsistencies might lead to odd results.

log %>%
    fix_resource_inconsistencies()
## *** OUTPUT ***
## A total of 4 activity executions in the event log are classified as inconsistencies.
## They are spread over the following cases and activities:
## # A tibble: 4 × 5
##   patient handling              handling_id complete start
##   <chr>   <fct>                 <chr>       <chr>    <chr>
## 1 401     Check-out             2630        r1       r5   
## 2 401     Registration          401         r7       r6   
## 3 401     Triage and Assessment 901         r6       r5   
## 4 401     X-Ray                 1685        r1       r7
## Inconsistencies solved succesfully.
## # Log of 10 events consisting of:
## 1 trace 
## 1 case 
## 5 instances of 5 activities 
## 5 resources 
## Events occurred from 2018-01-24 19:26:12 until 2018-01-29 13:25:13 
##  
## # Variables were mapped as follows:
## Case identifier:     patient 
## Activity identifier:     handling 
## Resource identifier:     employee 
## Activity instance identifier:    handling_id 
## Timestamp:           time 
## Lifecycle transition:        registration_type 
## 
## # A tibble: 10 × 7
##    patient handling   employee handling_id registration_type time               
##    <chr>   <fct>      <chr>    <chr>       <fct>             <dttm>             
##  1 401     Registrat… r7 - r6  401         start             2018-01-24 19:26:12
##  2 401     Registrat… r7 - r6  401         complete          2018-01-24 22:02:51
##  3 401     Triage an… r6 - r5  901         start             2018-01-25 04:32:05
##  4 401     Triage an… r6 - r5  901         complete          2018-01-25 13:41:02
##  5 401     X-Ray      r1 - r7  1685        start             2018-01-27 14:03:44
##  6 401     X-Ray      r1 - r7  1685        complete          2018-01-27 17:10:26
##  7 401     Discuss R… r2       2135        start             2018-01-29 09:07:13
##  8 401     Check-out  r1 - r5  2630        start             2018-01-29 12:05:20
##  9 401     Discuss R… r2       2135        complete          2018-01-29 12:05:20
## 10 401     Check-out  r1 - r5  2630        complete          2018-01-29 13:25:13
## # ℹ 1 more variable: .order <int>

Read more:

Create Logs Adjust Logs Public Logs Xes Files Inspect Logs Data Quality

  1. Currently both eventlog and activitylog are supported by the packages bupaR, edeaR and processmapR. The daqapo package only supports activitylog, while all other packages only support eventlog. While the goal is to extend support for both to all packages, you can in the meanwhile always convert the format of your log using the functions to_eventlog() and to_activitylog().↩︎

Copyright © 2023 bupaR - Hasselt University