Journey Discovery - User Guide
- Pete Trevitt
Journeys Discovery should only be run after the correct data has been collected and approach been agreed.
Please review Journey Discovery - Introduction and Considerations
Journey Discovery can be accessed from the Navigation panel on the left of the screen which links to the Journey Discovery Homepage
Journey Discovery Homepage
The Homepage displays all existing Journey Discovery that have been run. The display is based on the components of the Journey Discovery process. Each one is given a unique Id rather than a name, and each on can then be recognised by the configuration it used to execute.
Persistence and Saving
Once initially created, Journey Discovery executions auto-save in the last known state. States of execution can be seen on the Journey Discovery homepage as follows
Initialised - New Journey Discovery created but with no configuration
Prepared - A date range has been selected and a data set prepared
Calculated - The Micro Journey discovery process is compete but results have not been scored or uploaded to the result screen
Uploaded - Journey Discovery process is complete and results can be viewed
IMPORTANT NOTE: A Journey Discovery is based on its defined components, related to execution of the process against a specified Event Stream/Target.
Each execution of a Journey Discovery process will persist, save and re-use its component parts which are:
Date Range - Selects the data set that is built in the “preparing” section and will persist until the Date Range is changed.
Execution Parameters - Available parameters to guide the machine learning function and will persist until edited.
Data Transform - the version of the Data Transform set and fixed at start of process. This cannot be changed once created. If a later version of a Data Transform is required, a new Journey Discovery must be created.
Location Map - the version of the Location Map set and fixed at start of process. This cannot be changed once created. If a later version of a Location Map is required, a new Journey Discovery must be created.
Journey Ranking - Ranked list of Journeys created via Location Map, in terms of required focus for scoring and will persist until the Ranking order is changed.
Creating a new Journey Discovery
Open new Journey Discovery using the + icon and begin preparation stage
Select and Event Stream
Each Journey Discovery is tied to a single Event Stream for its lifetime. Select the required Event Stream from the drop down
UI will display currently published version of Data Transform & Location Map
The version is tied to the Journey Discovery for its lifetime
Provide Date Range selection
This is the data that will be used for the Journey Discovery
Date Range will be dependant on length of Journeys
Do not simply start with all the data as a short cut. This will lead to excessive calculation times unnecessarily.
Note: Closing at this stage will not Save the instance of Journey Discovery
Clicking Save and Prepare will create the data set in the back end, and move the status to Prepared
User will manually move to the next stage of process which Journey Discovery Calculation.
Based on the Date Range selected and the number of events, this process could take a short while (possibly minutes)
On competition a tick box will appear at the top of the Data Preparation tab.
Journey Discovery Execution Parameters
Once the data set has been prepared, that data can be passed to the Journey Discovery Process.
NOTE:
The output of this stage is a collection of Journey Pattern Files that are not yet available for view in the UI, as the Locations must be Ranked first, ready for display
There are no set guidelines here as the Analyst must assess the length Micro Journeys they are interested in.
This Journey Discovery process may be repeated multiple times with different configurations, on the already defined data set.
This may be an iterative and learning process to the analyst as they go through.
Only the last saved configuration is available at any one time.
The following defines the scope of the parameter
DeDuplicate
This option toggles between two different views of the data to be analysed. When data is collected, there are instances where a user might F5 refresh, or potentially the tag is creating multiple events, and the data would show the same Location twice or more in succession. This depends simply on whether the user is interested in this occurrence of whether the duplicated Location be considered as one, or not
No - Repetition of the same Location is considered in the Micro Journey calculation
Yes - Repetition of the same Location is NOT considered in the Micro Journey calculation
Minimum Pattern Length
When running a Journey Discovery process, this defines the minimum number of Locations that will be considered as a Pattern - Default 3 i.e. A Pattern must contain 3 or more Locations. User should consider if 3 Locations would constitute a pattern. This will come with experimentation
Maximum Pattern Length
When running a Journey Discovery process, this defines the maximum number of Locations that will be considered as a Pattern - Default 10 i.e. Patterns will only be considered with a this value as maximum number of Locations.
Minimum Repetitions
When running a Journey Discovery process, this defines the minimum number of times a Patterns must occur be be considered as significant. Default 100 i.e. Patterns must occur ore than 100 times to be considered for display in Journey Discovery
At this stage in the process there are no visualisations, as the user need to move to the Ranking phase, but as documented above, the user can return and change configurations if required.
Clicking Save and Calculate will run the Journey Discovery process to find Micro Journey Patterns based on the defined config
User will manually move to the next stage of process which Journey Ranking
Based on the Date Range selected and the number of events, this process could take a while (possibly minutes)
On competition a tick box will appear at the top of the Journey Discovery tab
Journey Ranking
At this stage the user is presented with a list of Journeys that were created as part of the Location Mapping process. The Journeys are initially displayed in the left hand “Unsorted” column, and are dragged across to create a prioritised list in the right. This part of the process is optional, and can be repeated multiple times on the same pattern result, to give a different priority to the Discovery Result, as explained later.
This action will apply a descending integer score to Locations as follows:
Locations in the Payment & Billing Journey - score 5
Locations in the Outages Journey - score 4
Locations in the Support Journey - score 3
Locations in the Login Journey - score 2
Locations in no Journey - score 1
When executing the scoring phase, each Location in the Micro Journey Discovered will be given a score. The highest scoring Location will be used to give the Micro Journey a name, but also assign that Location to that Journey within the Journey Discovery.
Usage of Scoring
It is important to note that scoring does not change the patterns, it simply gives the user a chance to label the patterns into a group based on the Locations and how important they are.
Also that this is an optional stage that can be repeated, again noting that each re-score does NOT change the patterns themselves but simply allows the user to prioritise Journeys that they wish to focus on for the particular piece of analytics question.
No Journey Ranking - In this case the analyst will simply calculate this stage with no listed ranking of Journeys i.e. No Journeys are added to the right hand column.
Journey Ranking - In this case the user will drag Journeys to the right hand column to give the Locations in that Journey priority over the lower or non ranked
If the analyst is most interest in the Payment & Billing Journey for example, this would have the highest rank, and name the Journey accordingly, even when they contain Locations from other Journeys. If the user was more interested in highlighting the Outages Journey, then ranking that firs would label that micro Journey in the Outages Journey, even if it contained some Locations from Payment & Billing.
Scoring can be repeated multiple times on the same micro journeys to allow a different emphasis on the Journeys as part of the output .
Once the scoring process is complete, the results can be viewed. The user will navigate to the Discovery Result tab.