Possible to filter on one dataset based on the results of another dataset in a page?

user018766
user018766 βšͺ️

I have a user table and an activities table. I want to track the activities over time, but since there are too many activities, if I include a user details for every activity, the dataset will be too large. I want to be able to filter the activities by features of the user. And the features are contained in the user table.

Eg:

User table has each row corresponding to a user and a lot of columns including ID (user id).

Activities table has only two columns, (1) ID, this is user id and the primary key of the first table. (2) Activity

By filtering on a column in the first table, country for example, is it possible to then have it tie up the IDs corresponding to the country that I have filtered on to the activities in the other table? So that when I select a particular country, only the activities of that country are shown?

Thanks in advance!

Best Answer

  • GrantSmith
    GrantSmith Indiana 🟀
    Accepted Answer

    Hi @user018766

    Currently no, you can't conditionally filter a dataset based on values from another dataset with different column names.

Answers

  • MarkSnodgrass
    MarkSnodgrass Portland, Oregon πŸ”΅

    As long as the column name that your clicking on exists in the other dataset, it will filter for you. That means if you click on country name, for example, country name would need to be in the other dataset. You also need to turn on interaction filter on your page filter for the card to card filtering to work.

    I would also point out that Domo is built to handle millions and millions of rows. You can have separate datasets that have been uploaded and you can join them in Magic ETL (rather than before you upload them) and Domo processes the data very quickly. This would allow you to have one dataset to use and make the card to card filtering easier for you.

  • user018766
    user018766 βšͺ️

    Thanks. It doesn't answer my question though, unless your answer is 'no, it isn't possible'. I am aware that we can filter based on two shared columns. But I am trying to filter values on one dataset by using the filtered output on another.

    About it not taking long, I'm not sure. I have an operation where a million rows are collapsed into 5 million and that takes 45 minutes. And the dataset that I am currently working on has much more data.

    I will not be able to filter the way I want will I? Klipfolio for example has this feature, would be nice if Domo had it too.

  • jaeW_at_Onyx
    jaeW_at_Onyx Budapest / Portland, OR 🟀

    @user018766 "if I include a user details for every activity, the dataset will be too large" how do you know that statement is true?


    by too large do you mean "too many columns" or "too many rows"? because from the Domo architecture perspective, neither statement is true.


    "I have an operation where a million rows are collapsed into 5 million and that takes 45 minutes" are you using ETL to transform a 5M to 1M? if so which engine are you using? because 45 minute + Magic ETL 2.0 should be fine. And ideally if all you're doing is Joining datasets, consider Dataset Views (it's a beta talk to your CSM) or Fusions (part of General Availability) the upside of views is they execute in the database layer and update virtually instantaneously when the underlying data updates.

    "By filtering on a column in the first table, country for example, is it possible to then have it tie up the IDs corresponding to the country that I have filtered on to the activities in the other table?Β " ... just JOIN your dataset.

  • user018766
    user018766 βšͺ️

    @jaeW_at_Onyx Thanks, any idea if dataset size will have an effect on time taken to filter in the dashboards? (i.e. Time between clicking 'Apply' in the filter pane and all the charts finish loading). The dataset size should be around 1 billion rows and 30 or so columns with mixed datatypes.

  • jaeW_at_Onyx
    jaeW_at_Onyx Budapest / Portland, OR 🟀

    @user018766 you're in luck, i've done a video on user interactions with large datasets :P

    with 1B rows you do want to start making good decisions on how you structure your data. I still recommend joining your data using dataset views, just limit the columns you're including. 30 columns is fine.


Sign In or Register to comment.