can anyone give me a perfect documentation for time series to understand the things easy and clear or any track with steps of how to create I used marketplace’s but in some parts it tough to reverse engineer
so, anyone who has some idea on it please help
Did you start with https://www.palantir.com/docs/foundry/time-series/time-series-overview ?
Might you point out what is unclear in particular or what confuses you, so that we can be more helpful there ?
Like while creating tsp we need to use sync, for that we need to create a sync what dataset need to be used to create sync like we need to use the dataset which was used for ontology what is the structure for of the cols need to been in so that we can create a sync
Like in pipeline builder what transformations needs to be done to achieve the structure we need
I know this seems a little confusing if i was wrong somewhere please correct me
give me correct flow of how it needs to be done and also how to use them in quiver
I get an understanding but not able to find a correct path
If Any existing Tracks Like things exist that would be helpful
You can look at the examples available in the “Examples” application in platform.
This should give you more insight as to how to set-up such capability.
From a more generic explanation, this is what you need:
You need on one side to have “series” of points stored in a time-series sync.
On the other side, you need an object with a property that is the same value as the “series” stored in the time-series sync. So that it knows “which series” to pick up from this time series sync.
From there, the rest of the platform should know what to do with that. You need to reference the time-series sync itself in the “capabilities” tab in Ontology Manager (OMA).
@VincentF I understand series sync created and used in capabilities tab needs to be same
But to create a sync we need specific columns like Series timestamp and values (which can be created by unpivoting the cols which are used for series but here to do that we need some dataset (consider A)) and to create a TSP we need a dataset like to create an object (it also needed to be same dataset which is A?)
here u see backing dataset for TSP creation was airports
if you see the sync the data set used was different
could you please explain why the specific dataset was used
And also for creating a sync we need some specific cols for we need to do unpivoting (according to my understanding )
similarly if you see there were some cols in Airport object type which has its type as timeseries.
to achieve that we can directly change the type or we need to achieve some specific format of col to do that
and also After adding time sync in capabilities should we need to do any changes to the properties
I tried My best to give my view hope you understanding please give me the right flow
Its been a week I am stuck here
If you get a closer look at the solution design diagram above and the docs, those are not the same dataset.
You have one dataset (let’s call it A) with:
- Serie ID = any string
- Timestamp of the event = timestamp type
- Value = can be numerical (double, integer, etc.) or a string (e.g. “READY”, “ERROR”, etc. that shows the state of something,t technically called “categorical”)
This dataset should be synced as a time-series sync.
Example:
| serie_id (string) | timestamp (timestamp) | value (integer or double) |
|---|---|---|
| A | 2026-04-15T09:00:00Z | 1 |
| B | 2026-04-15T09:00:00Z | 1 |
| A | 2026-04-16T09:00:00Z | 2 |
| A | 2026-04-17T09:00:00Z | 3 |
| B | 2026-04-16T09:00:00Z | 10 |
| B | 2026-04-17T09:00:00Z | 20 |
You should then have another dataset (let’s call it B), with, at least:
- A primary key = that’s for object indexing, it needs to be unique. Highly advised to be string type
- A column containing the relevant “serie ID”, string type. They should match the ones of the first dataset.
You should configure this second dataset as a backing dataset for your object type.
Example:
| pk (string) | serie_id (string) |
|---|---|
| pkA | A |
| pkB | B |
How to obtain those two datasets can vary a lot. You can do this by indeed unpivoting, or via other means, being it in Pipeline Builder, Code repository, etc.
There is nothing hidden from there. Time-series sync, etc. won’t unpivot for you.
Let me know if something is unclear on the above, or if you have more precise questions/blockers.
ok understood pretty well @VincentF
so now how to create the properties that are time series
so that we can use them in different application
and while creating a object type whats the procedure like
in capabilities we will add sync
and then what to do like we need to use them in properties or just chanfe the base type to series
and how they can be used in quiver or workshop(breif it a bit more use can use the example in marketplace which i have used here so that we can be on same track if possible do that with previous example as well) market place was Building with Time Series Data
- You add a property of type “Time series” ( you can see the “base type” field on the right).
- Just below this field, there is a “Capabilities tab” link you can click on. This will lead you to the Capabilities tab.
- Once in the Capabilities tab, you can select the backing TimeSeries sync (the inference of “what series type is in the property of this particular object instance” to “which time series from the time series sync is related to” is done automatically.
- As long as the serie id (the actual string populating the property) is the same, this will work out of the box. Aka: If you load the object type in Object Explorer, Insight, Workshop, Quiver, etc. the property of type “time series” will show up as a “time series line” (the same way that a string property would be presented as a string, etc. no need for more configuration from there)
In the picture you share, you cannot edit the object type. So you won’t be able to do it, on this one, in particular. But you can do it on an Object Type on which you are editor/owner.
Yeah that’s the problem i was not able to edit them
Can you do one thing as you said like there need needs to be a series id that has to be related/common in the above marketplace mentioned can you tell how they are connected
like what is common series id
I want to know if there is some example like airplanes what type to connection we need to establish between sync and object type so they are connected
like you say we have this things and we have to acheive sync from this(dataset let i be A use some example like temparatue or dalay just usecase will be helpful) and object type (datasetB ) and the connection betwwen them is this (series id ) to achive something from that we need to do this so that we can acheive
You need to create an object type manually as you can’t edit this one, installed to back the examples.
You need to look at this for the Object Type setup: https://www.palantir.com/docs/foundry/object-link-types/create-object-type#create-a-new-object-type
And for the timeseries part: https://www.palantir.com/docs/foundry/time-series/create-or-select-ts-ot/
sure mate
Thanks for the adivce
Great talking to you