Survey banner
Share your feedback on the Dataiku documentation with this 5 min survey. Thanks! TAKE THE SURVEY

Checks method values update

Solved!
sunith992
Level 3
Checks method values update

Hi,

can anyone please answer for the below issue.

the Metrics in the status tab is not updating even when the dataset is modified/built. 

i am using a Python notebook to update/rewrite the a dataset which already had made setup in such a way that whenever the Dataset is updated/built  it supposed to Auto completes the metrics which we selected, below  is the option located in it.

Auto compute after build:

unfortunately it's not updating the metrics tab i need even after i made these settings, i am again re-running the metrics by selecting run and going back to the metrics tab.

also,  when i click refresh Dataset/partitions its consuming huge amount of time and sometimes not responding.

0 Kudos
2 Solutions
SarinaS
Dataiker

Hi @sunith992,

It would be helpful to see screenshots of your flow and where this dataset is in your flow. Is it possible that this is an input dataset to your flow that you are looking at? (i.e. a dataset that doesn't get built?) 

If this is an output dataset, can you clarify how exactly you are building the output dataset (i.e. through a scenario step, clicking "run" in the recipe tab directly etc.). 

If you are using a Python notebook to update the dataset, then indeed the dataset is not getting "built". Running code from a notebook does not perform a dataset build. You must instead run a build step from the flow or a scenario to actually build a dataset in order to perform a dataset bulid .

In order to create a build flow for the dataset, you will likely want to convert your notebook into a Python recipe that builds the output dataset. Then, if you schedule this build in a scenario or run a build of this dataset from the flow, the metrics should be automatically computed. 

You can convert your notebook to a recipe by clicking on the "Create recipe" option from a notebook:

Screen Shot 2022-10-10 at 5.59.20 PM.png

Your Python code must have the dataset you are computing metrics for as the output dataset so that every time you run this Python recipe, it will build the output dataset and can correspondingly compute metrics. 

Let me know how this goes and if you have questions after running a build of the dataset.

Thanks,
Sarina 

View solution in original post

0 Kudos
sunith992
Level 3
Author

Hi Sarina,

Thanks for the response, now i am good with the above answer.

also would like to express that in the Dataiku documentation, nowhere explicitly mentioned about  the python notebooks where these notebooks are not allowed for building the dataset computation metrics even though its updated/replaced in the notebook itself.

View solution in original post

2 Replies
SarinaS
Dataiker

Hi @sunith992,

It would be helpful to see screenshots of your flow and where this dataset is in your flow. Is it possible that this is an input dataset to your flow that you are looking at? (i.e. a dataset that doesn't get built?) 

If this is an output dataset, can you clarify how exactly you are building the output dataset (i.e. through a scenario step, clicking "run" in the recipe tab directly etc.). 

If you are using a Python notebook to update the dataset, then indeed the dataset is not getting "built". Running code from a notebook does not perform a dataset build. You must instead run a build step from the flow or a scenario to actually build a dataset in order to perform a dataset bulid .

In order to create a build flow for the dataset, you will likely want to convert your notebook into a Python recipe that builds the output dataset. Then, if you schedule this build in a scenario or run a build of this dataset from the flow, the metrics should be automatically computed. 

You can convert your notebook to a recipe by clicking on the "Create recipe" option from a notebook:

Screen Shot 2022-10-10 at 5.59.20 PM.png

Your Python code must have the dataset you are computing metrics for as the output dataset so that every time you run this Python recipe, it will build the output dataset and can correspondingly compute metrics. 

Let me know how this goes and if you have questions after running a build of the dataset.

Thanks,
Sarina 

0 Kudos
sunith992
Level 3
Author

Hi Sarina,

Thanks for the response, now i am good with the above answer.

also would like to express that in the Dataiku documentation, nowhere explicitly mentioned about  the python notebooks where these notebooks are not allowed for building the dataset computation metrics even though its updated/replaced in the notebook itself.