Community Discussions

We are proud to announce the publication of our latest research on the application of Topological Data Analysis (TDA) for Condition-Based Monitoring (CBM) of wind turbines. This new study will be presented next week at an international conference focused on equipment health and prognostics in Prague https://phm-europe.org/.Abstract: Our research investigates how TDA, a sophisticated branch of data analysis, can enhance the monitoring and maintenance of wind turbines. By analyzing complex datasets obtained from standard vibration sensors in turbine gearboxes, we identify patterns, anomalies, and trends that are often undetectable using traditional methods.Key Highlights:Data Source: gearbox vibration data, collected from a wind park in Norway, data contextualised in CDF. Methodology: Conversion of time series data into multi-dimensional point clouds through time-delay embedding Analysis Tools: Utilization of topological methods, including persistent homology Indicators: Key health indic

I am using the Python SDK endpoint to aggregate instances and passing an advanced Filter to the filter argument, i.e.:client.data_modeling.instances.aggregate(…, filter = Filter(...))But Nested filtering is not supported. It seems like the aggregate function only works on flat tables.So I am considering to create a solution data model with only flat tables. I am considering to split up the data over multiple containers in order to support the number of indexes I would like to have. My question is whether splitting up over multiple containers negatively affect performance at query time and how does this compare to trying to limit the number of indexes instead? 

hello:I want to associate files uploaded to CDF type pdf for example to my model, following this scheme:type Documentation { name: String # This is a CDF built in type file: File # this is the INVERSE relationship of the Pump.documentation # Pump.documentation <-> Documentation.relatedPumps relatedPumps: [Pump] @relation( direction: INWARDS, type: { externalId: "Pump.documentation", space: "Space_PabloPalacios" } )}but when I select the uploaded file it only shows me jpg, any ideas?

On behalf of Celanese, I would like to report an issue we encountered with the SDK Data Modeling instances aggregate, where the filter format changed from yesterday, 06/11, to today, 06/12. Until yesterday, the "equals" filter worked, but today it broke all the calls that used it and had to be modified to the "in" format, which did not work correctly until yesterday.Another bug to report in the aggregate call is that when an array of group bys is sent, the return structure has also changed, no longer receiving a structure with externalId and space, but a string concatenating the two in the format: "space:externalId”, and its nomenclature changed from "GroupByName" to "GroupByName.__direct_relation". 

How to PAN the 3D Model left, right, up, down

Hi! I’m trying to create a Power BI Dataflow using CDF timeseries data. The same data is retrieved absolutely fine when using Power BI Desktop and Power Query inside the Power BI Desktop app, but when I try to use exactly the same query as a Dataflow instead, there is a warning about: “The query “Timeseries” contains columns with complex types that cannot be loaded. Some data is retrieved, but it’s incomplete with big parts of it missing. Looks like the CDF connector is not able to retrieve the data correctly when a Dataflow is used instead of a normal Power BI Desktop dataset. Is there a way to make Dataflow also compatible with the CDF connector?  

I want to urge the developers to write a documentation that tells us how to use the available functionality.Specifically I am looking at the filtering functionality. Let me explain my use case: I need to query CDF for a lot of time series. The unfiltered query returns a TimeSeriesList of 32k TS. I only need a subset of these, and want to perform the filtering on the CDF side. I want to fetch the subset of TSs that contains any of the following strings in the name attribute (yes, the dashes and underscores included)-PT- -PI- -PIC- -PIT-Can someone please explain how to do this? Search does not match the dashes and underscores, so I think it performs a fuzzy search. But it is not explained in the docs.ContainsAny appears to only work when the property of interest is an array. So I cannot use this filter to check if a string property contains any of a list of search strings. But this limitation is drastically undercommunicated in the docs. After reading the docs several times it looks lik

I am facing slow response issue while working with Jupyter Notebooks. Also, when I restart the kernel it takes almost 5-7 minutes to get it started.Is there any standard method to overcome this?   

Hello team,We are trying to use the data modeling query instances SDK method.client.data_modeling.instances.query(query) our query is:query { "with": { "Min_15_50ts": { "nodes": { "filter": { "and": [ { "hasData": [ { "type": "view", "space": "QA_Test_Space", "externalId": "Min_15_50ts", "version": "3a62f20104e445" } ] }, { "in": { "property": [ "QA_Test_Space", "Min_15_50ts/3a62f20104e445", "entity" ],

The camera control commands seem to change over time and I couldn't identify why that happens in our app, though I was able to see a similar behavior in CDF 3D Scenes.The default control  when dragging with mouse's left button is to describe a rotation of the camera, but keeping the focus point in place (Gif 1). Sometimes, however, the same mouse movement causes the camera position to change describing an arc-like movement (Gif 2). While I think that “static” mode is using the “Orbit” control, the “arc” mode does not look like the other option, “Fly” control, because on the Fly control the camera position (shown at the top-right) does not change, but in the “arc” mode it does.When I click the "Home" button, a red focal dot in the center of the screen vanishes and it goes to the “arc” mode. When I click the "Fit View" button, the red dot reappears and camera commands are back to the “static” mode. There is nothing particularly wrong with either “static” or “arc”, but I'd like to keep th

Can someone please suggest, how we can add additional python libraries in CDF jupyter notebook?

The information I found here is contradictory: Cognite HubSpecifically, I am interested in Aker BioMarine’s tag data on CDF (example shown below). I did not check the documentation, since I prefer to hear it here, “straight from the horse’s mouth”, if you will.Questions:Is the same time zone used for all of our assets, regardless of location of the geographic location of our manufacturing site? Is the time stamp preserved for time series downloaded using, e.g., Python / Jupyter notebook? Anything else I should be aware of?Thank you very much!

I’m currently writing my Master thesis in collaboration with Aize (represented by V. Flovik), where I’m trying to use Generative adversarial Networks (GAN) to generate synthetic time series data. To do this, I’m using data from Cognite OID. I’ve explored the data in the previous semester, and I figured that one of my main challenges is to find longer periods of data that are “Normal”. Meaning periods without to large irregularities and without too many missing data points. Because even when I aggregate the samples over 30 seconds, there are still samples missing in some periods. Some of these periods are hours long and can’t really be filled in without messing up the temporal development of the data. I can’t find any information about the cause of these periods either.The time series I’ve been looking at are:I’m no expert in process engineering, and I only chose these 9 sensors because another master thesis from 2019 had used these sensors in their project. I therefore reach out here t

We need to understand if Cognite endpoints offer a way to do few shot learning (custom prompt)

I am unable to generate client code for the CDF openapi spec at https://api-docs.cognite.com/20230101/ using deepmap/oapi-codegen for Go, seemingly because the spec contains some errors. Do anyone know of a way to generate an api-client in go for the CDF-api spec? Both commands below rely on having openapi spec downloaded to the current directory from https://api-docs.cognite.com/20230101/ Testing the spec for errors with redocly/cli: docker run --rm -it -v $PWD/swagger.json:/swagger.json redocly/cli lint /swagger.jsonFor me this results in “Validation failed with 25 errors and 382 warnings” Generating the client library using deepmap/oapi-codegen: go run github.com/deepmap/oapi-codegen/v2/cmd/oapi-codegen@v2.0.0 --config oapi-codegen-config.yaml swagger.jsonwith the this config file:package: "cdf_api"output: "cdf_api.gen.go"generate: models: true client: truecompatibility: circular-reference-limit: 100For me this results in the message “error generating code: error creating operat

Hello Cognite Community,We are thrilled to invite you to join our Early Adopter Program focused on exploring what CDF usage metrics can help you evaluate the value of Cognite Data Fusion (CDF) for your operations.Why Join?Uncover Key Metrics: Help us identify the most impactful usage metrics that demonstrate CDF's value for your company. Drive Improvement: Your feedback will guide us in refining how we measure and communicate the benefits of CDF. Exclusive Engagement: Be among the first to provide insights and shape the future of CDF usage analytics.How to Participate:Comment Below: Share your experiences and thoughts on which metrics could best capture CDF's value for your operations. Direct Message Us: Prefer a more private discussion? Send us a direct message to discuss your usage metrics in detail. Like This Post: If you find this relevant, please like this post to show your interest, and we’ll send you an invite to join the program.By participating, you’ll play a key role in enhan

Hi team,This is what I am trying to achieve:I have written a function to create a new time series (in case it is not already created) and insert data points into it.The newly created timeseries needs to be added to a solution data model I have created.type ESPInputs @view(space: "workflow-sdm-spc", version: "1_0") {    well:Well    pump: ESP,    operatingFrequency: TimeSeries,    voltage: TimeSeries,    pumpIntakePress: TimeSeries,    pumpDischargePress: TimeSeries,    motorWindingTemp: TimeSeries,    current: TimeSeries,    vibration: TimeSeries,    torque: TimeSeries,    pmm: TimeSeries,    operatingFrequencyPmm: TimeSeries}The instances under this view are already present. The newly created time series need to be references to the property pmm that are filtered via well property (I have well external Id and pmm time series external id available with me in the function).I think I need to update the instance in the apply instance as per the documentation, but I am unable to find an up

The navigation around a busy canvas is awkward. For example having to switch from Select mode to Grab mode to move around is not intuitive. Relatedly, when in grab mode and you hover over a chart for example, you would expect to be able to click the chart and then interact with it. You cannot. You must switch to select mode to do that.I suggest that if the mouse is in “white space” then grab mode is in effect by default. If the mouse is over an object then it should switch to select mode automatically. 

I type in a comment in the comments pane and there is no way to submit the comment. I press enter and nothing happens. There is no button to submit. When I close the comments pane and reopen, the comment text has been deleted. Am I missing something?

Data Explorer (all resources tab) inconsistencies. Login to CDF Navigate to Data Explorer screen. Search the Asset name e.g: 'Inlet separator' in search field. Click on the All-Resource tab to open. Expected Results: All resources screen data should be relevant to the Asset which user searched in search tab.Observed Results: All resources page does not point to relevant relationships – better to focus on direct linkages.

Hyperlink in CDF is not navigate to right screen for the provided hyperlink. User will navigate to CDF by default screen not the intended hyper linke.g: https://oq.fusion.cognite.com/oq-test/explore/search/timeSeries?cluster=westeurope-1.cognitedata.com&env=westeurope-1&journey=timeSeries-114626196540160&q=separator  

The number of assets shown as linked to the file is not correct. The system is wrongly showing the number of assets linked in the file.And also Linked Asset to an asset count is wrong and same issue for linked files, linked time series, Linked events. These linked is different than directly linked items. (A) Linked (Files): this count means all decedent children of a given asset. (B) Directly linked (files): this count is the number of files that are connected directly to the asset. (C) The file count on top: This count may vary since there might be duplicated IDs amongst the different types - the total number is a distinct value. The linked assets are not relevant to the Asset which user try to refer and these Linked assets will not linked to the Asset.Directly linked assets are linked to the assets which user try to refer.User don't want to see linked asset tab in CDF because its doesn't have a meaningful assets linked to it, however Directly linked asset have meaningful tab and same

CDF - Filter option is not working as expected under common filters at Data Explorer screen. Login to CDF Click on Data Explorer tab in CDF menu bar. Click on Files tab in right side of the panel. Set Data set as 'src:006:documentum:b60:ds under Common filters in left side of the screen. Select the check box ‘Before’ under common filters in left side of the penal. Click on the Calendar icon and set data as (e.g.) '10-01-2023' Expected results: Document ‘Amarjeet_Test_DT.docx’ should not display in results window because its created after the set date.Actual results: Document Amarjeet_Test_DT.docx is displaying in CDFNote Issue exists for all Date filters like Created time, updated time with Before, After, During in CDF, user want to know what date is used for filtering the documents in CDF with these filters.

Is the following use case a good fit for Data Workflows?Cognite Function that is reading data from Time Series and writing "event-like" data into Data Modeling continuous processing for metric type A ideally an execution every minute, but more importantly no overlapping executions (i.e., if execution takes longer than a minute) passing state from one to another function execution hourly processing for metrics type B Idea:2 Data workflows:Hourly execution of trigger Cognite Function to calculate metric B Daily execution of a workflow that triggers Cognite Function that recursively outputs a dynamic task, which calculates metric A. The Function outputs information for the next execution run (i.e., timestamp for next execution, state, and other info). At the end of the day the dynamic task would not output a timestamp for the next execution and workflow would complete. The execution trigger could also be daily.

Hello, When I tried to run the DBExtractor, I get the following error: “polars\_cpu_check.py:232: RuntimeWarning: Missing required CPU features.The following required CPU features were not detected:    avx, avx2, fmaContinuing to use this version of Polars on this processor will likely result in a crash.Install the `polars-lts-cpu` package instead of `polars` to run Polars with better compatibility.Hint: If you are on an Apple ARM machine (e.g. M1) this is likely due to running Python under Rosetta.It is recommended to install a native version of Python that does not run under Rosetta x86-64 emulation.If you believe this warning to be a false positive, you can set the `POLARS_SKIP_CPU_CHECK` environment variable to bypass this check.”After doing some googling, I was able to install the polars-lts-cpu package referenced using python, but I got the same error. I’m not sure how to make the extractor reference the polars-lts-cpu package when it runs. See attached screenshot.The extractor i