create_prediction: Create a prediction on a specified experiment_version

Description

Create a prediction on a specified experiment_version

Usage

create_prediction(
  experiment_version_id,
  dataset_id = NULL,
  folder_dataset_id = NULL,
  confidence = FALSE,
  best_single = FALSE,
  model_id = NULL,
  queries_dataset_id = NULL,
  queries_dataset_content_column = NULL,
  queries_dataset_id_column = NULL,
  queries_dataset_matching_id_description_column = NULL,
  top_k = NULL
)

Arguments

experiment_version_id

id of the experiment_version, can be obtained with get_experiment_version_id().

dataset_id

id of the dataset to start the prediction on, can be obtained with get_datasets().

folder_dataset_id

id of the folder dataset to start prediction on, can be obtained with get_folders(). Only usefull for images use cases.

confidence

boolean. If enable, confidence interval will be added to predictions.

best_single

boolean. If enable, best single model (non blend) will be used for making predictions other wise, best model will be used unless if model_id is fed.

model_id

id of the model to start the prediction on. If provided, it will overwrite the "best single" params.

queries_dataset_id

id of the dataset containing queries (text-similarity).

queries_dataset_content_column

name of the content column in the queries dataset (text-similarity).

queries_dataset_id_column

name of the id column in the queries dataset (text-similarity).

queries_dataset_matching_id_description_column

name of the column matching the id (text-similarity).

top_k

number of class to retrieve (text-similarity).

Value

list - parsed prediction information.