Cognite Model Hosting Documentation


To install this package run the following command

pip install cognite-model-hosting


cognite-model-hosting is an open-source library containing utilities for working with data from the Cognite Data Platform (CDP) in the Model Hosting environment. Working with the data is split into two parts;

  1. Specifying data, such as time series and files, using Data Specs
  2. Fetching the described data using the Data Fetcher.

Data Specs are a collection of classes used to specify data in CDP. These specs are organized in a hierarchical way, so that they can be collected in a single object, sent easily around and passed to an instance of a DataFetcher. The Data Fetcher is then used to retrieve the specified data from the platform.


from cognite.model_hosting.data_fetcher import DataFetcher
from cognite.model_hosting.data_spec import *

id1 = ...
id2 = ...
ts1 = TimeSeriesSpec(id=id1, start="3d-ago", end="now", granularity="1d", aggregate="average")
ts2 = TimeSeriesSpec(id=id2, start="3d-ago", end="now", granularity="1d", aggregate="max")
ds = DataSpec({"myts1": ts1, "myts2": ts2})

data_fetcher = DataFetcher(ds)
df = data_fetcher.time_series.fetch_dataframe(["myts1", "myts2"])