Hi guys, We have a very complex data model and whe...
# gd-beginners
Hi guys, We have a very complex data model and when mapping this on the LDM, I get a lot of
Copy code
The model contains dataset relations that allow alternative paths. This will produce inconsistent results. Update the dataset connections to prevent ambiguous paths.
warnings. Is there a technique to avoid these but still make our reports makes sense? For example, I can see the auto-hiding "unrelated" items in the analytics editor actually hiding things we want to select as it seems LDM getting confusing. Any advice/ best practice?
Hi Dilaksha, Yes, this kind of relations are still valid. However, then our analytical engine randomly selects between the available paths in any computation and that might produce non-deterministic results. Some hints can be found in this article or in this community post. And since you are working with complex model, I’d also recommend to check our University courses: https://university.gooddata.com/data-modeling-essentials https://university.gooddata.com/advanced-data-modeling
Not seeing preferred objects in your AD catalog, might be part of this “issue”, either there are too many alternative path, or there is no connection between the datasets at all.
Hi @Ivana Gasparekova thanks, I will check out the courses. What did you mean by preferred objects? if you see my LDM and master workspace, I created a analysis called "Ivana" and problem is I want some fields from the dataset "rental product inventory" but it is not available for me to choose. There surely too many paths problem exists, just want to know how to resolve these as the relationships are correct
Oh sorry, I meant the objects(attributes, facts, dates) which are being recognized as “unrelated”, but which you’d like to use in your analytics together.
Please understand that unless you invite our support user to your project/workspace, we are unable to access it from our end. You might solve it by adjustments of your metrics and “navigating” the engine through the right path. This article is using pictures from the older UI - PixelPerfect, but it might be still useful for your situation. https://help.gooddata.com/classic/en/dashboards-and-insights/maql-analytical-query-language/maql-use-cases-and-tutorials/explicit-lifting/
The only other option on my mind would be to optimize the model.
Hi, I would higly recommend this article as well: https://community.gooddata.com/data-models-kb-articles-46/logical-data-model-basic-rules-of-data-modelling-176 specifically point 5 there which deals exactly with this topic. (Actually you might fond this whole series of articles about data model useful).
🙌 1