Application upgrade tracking

Coni

I am tracking an application update on devices. I have a dataset that looks something like this. I would like to:

1. Count # of devices that upgraded to the latest app (Column App1). Ex. is 1002 and 1004 that has 2 versions, I only need the latest which is in column App1.

2. Count # of devices with app1 and app2

3. Count devices # of devices with various app2 versions

4. Ability to track these changes by weekly. I currently do not have a date column.

Serial Number	App 1	App 2
1000		2.8
1001		2.8
1002	3	2.6
1003		2.7
1004	3	2.6
1005		2.7

Thank you in advance.

GrantSmith

In order to track the weekly stats you'd need to either use a dataflow to add in the timestamp and schedule it to run once a week and have it append on the output or alternatively use a dataset copy connector and set it to append to automatically copy the original dataset with the _BATCH_LAST_RUN_ timestamp field.

As for your other questions:

1 - How are you determining if they're on the latest app? Is that if App 1 is populated? IF that's the case you can just get a Count of App 1 to count how many serial numbers have it installed. This is assuming you have one record per serial number. If not you'd need to do a distinct count in a beast mode:

COUNT(DISTINCT CASE WHEN `` IS NOT NULL THEN `Serial Number` END)

2- You can use a conditional SUM within a beast mode:

SUM(CASE WHEN `App 1` IS NOT NULL and `App 2` IS NOT NULL THEN ` ELSE 0 END)

3 - Are you wanting to know which devices have multiple versions of App 2 installed or how many different App 2 version are installed across all versions?

GrantSmith

In order to track the weekly stats you'd need to either use a dataflow to add in the timestamp and schedule it to run once a week and have it append on the output or alternatively use a dataset copy connector and set it to append to automatically copy the original dataset with the _BATCH_LAST_RUN_ timestamp field.

As for your other questions:

1 - How are you determining if they're on the latest app? Is that if App 1 is populated? IF that's the case you can just get a Count of App 1 to count how many serial numbers have it installed. This is assuming you have one record per serial number. If not you'd need to do a distinct count in a beast mode:

COUNT(DISTINCT CASE WHEN `` IS NOT NULL THEN `Serial Number` END)

2- You can use a conditional SUM within a beast mode:

SUM(CASE WHEN `App 1` IS NOT NULL and `App 2` IS NOT NULL THEN ` ELSE 0 END)

3 - Are you wanting to know which devices have multiple versions of App 2 installed or how many different App 2 version are installed across all versions?

Coni

Thank you for your response. I will add a timestamp to my dataflow. I wanted to get a count of how many devices have 2 verions of the app. And how many just have one or the other version.