Skip to content

Conversation

@jaidevd
Copy link
Contributor

@jaidevd jaidevd commented Jun 7, 2022

"Prophet" is available as a valid model class in MLHandler now.

Usage

Use the following spec:

mlhandler/prophet:
    pattern: /prophet
    handler: MLHandler
    kwargs:
      model:
        class: Prophet
      xsrf_cookies: false

Download a sample dataset here

  • Training:
curl -X POST "/prophet?_action=train&target_col=y" -d @train.json -H "Content-Type: application/json"
# {score: 0.33} Returns the MAE score
  • Getting predictions against timestamps
curl -X POST "/prophet?_action=predict&target_col=y" -d @test.json -H "Content-Type: application/json"
# The timestamp column MUST be called "ds"
  • Getting forecasts for n periods in the future
# Get a month's forecast, assuming daily data
curl -X GET "/prophet?_action=predict&n_perids=30"

To be merged after #547

jaidevd added 8 commits May 11, 2022 13:17
Sklearn models are saved as files, transformers are saved in
directories. Earlier, the handler was doing disk I/O - this commit
ensures that the wrapper classes perform the I/O for the models
MLHandler now supports trainable named entity recognition.
The MLHandler template now has a UI for sentiment analysis with
transformers. A labelstudio interface is added for fine-tuning the model
with live annotations
"Prophet" is available as a valid model class in MLHandler now.
@sync-by-unito sync-by-unito bot changed the title ENH: MLHandler supports FBProphet for forecasting GRAMEX-205 ⁃ ENH: MLHandler supports FBProphet for forecasting Jun 7, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants