remove_question
DELETE /v1/scoring_models/{model_id}/questions/{question_id} removes a question from a scoring model, without calibrating it.
A question the calibrated score uses is removed with remove_questions when you calibrate instead. See Changing questions. Free.
Removes a question from a scoring model, without calibrating it. A removed question isn't asked or tried again.
- If it isn't calibrated yet, you can remove any question, as long as one is left.
- If it's calibrated, you can remove your own questions it doesn't use. A question the score uses is removed with
remove_questionswhen you calibrate, so the score is refit without it.
Path parameters
| Field | Type | Description |
|---|---|---|
model_idrequired | string | The scoring model's |
question_idrequired | string | The question's |
Responses
The scoring model, as get_scoring_model returns it.
| Field | Type | Description |
|---|---|---|
model_id | string | The scoring model's id, |
name | string | A few words saying what it predicts: the |
status | training, ready, failed |
|
criteria | string | The |
created_at | string | When the scoring model was created. |
calibrated_at | string or null | When it was last calibrated, or |
archived_at | string or null | When it was archived in the Finic Portal, or |
examples | integer | The number of examples it was created from, or first calibrated with; 0 until then. |
bad_rate | number | The share of those examples with |
scored | integer | The number of records scored with it. |
outcomes | integer | The number of those records with an outcome sent to calibrate. |
questions | array of object | The questions it uses, the strongest first. Your own questions it doesn't use are listed on its page in the Finic Portal. |
| Field | Type | Description |
|---|---|---|
id | string | The question's id, such as |
text | string | The question |
source | finic, user |
|
weight | number | How much it moves the score, from the latest calibration, or Finic's estimate if the scoring model isn't calibrated yet. Positive when a likely yes raises the score, negative when it lowers it; the larger, the more it counts. |
| Status | Type and when |
|---|---|
400 |
|
401 |
|
404 |
|
409 |
|
429 |
|
Every error has the same shape; see Errors.
Examples
curl -X DELETE 'https://api.finic.ai/v1/scoring_models/string/questions/string' \ -H "Authorization: Bearer $FINIC_API_KEY"{ "model_id": "scm_01M4XRC000RT2GZSKYPD8J86G9", "name": "Card-not-present payment fraud", "status": "ready", "criteria": "Card-not-present payments at an online store. Bad: the payment was charged back as fraud within 90 days.", "created_at": "2026-10-20T17:04:11Z", "calibrated_at": null, "archived_at": null, "examples": 0, "bad_rate": 0, "scored": 12, "outcomes": 0, "questions": [ { "id": "q003", "text": "This record is a card payment. Is the IP address in a different country from the billing address?", "source": "finic", "weight": 1.24 }, { "id": "q009", "text": "Was the account opened less than a week before the payment?", "source": "user", "weight": 0.93 }, { "id": "q006", "text": "This record is a card payment. Has this card paid this merchant before without a dispute?", "source": "finic", "weight": -0.71 } ]}