Sign in
hidden

Foundational Time Series Model (Beta)

This endpoint predicts the future values of a single time series based on the provided data. It takes a JSON as an input containing information like the series frequency and historical data. (See below for a full description of the parameters.) The response contains the predicted values based on the input arguments. Get your token for private beta at https://nixtla.io/free-trial?utm_source=nixtla.io&utm_campaign=/docs/api-reference

post/forecast

Request body

fewshot_stepsinteger nullable

Deprecated. Please use finetune_steps instead.

fewshot_loss'default' | 'mae' | 'mse' | 'rmse' | 'mape' | 'smape' nullable

Deprecated. Please use finetune_loss instead.

modelstring

Model to use as a string. Common options are (but not restricted to) timegpt-1 and timegpt-1-long-horizon. Full options vary by different users. Contact support@nixtla.io for more information. We recommend using timegpt-1-long-horizon for forecasting if you want to predict more than one seasonal period given the frequency of your data.

freqstring

The frequency of the data represented as a string. 'D' for daily, 'M' for monthly, 'H' for hourly, and 'W' for weekly frequencies are available.

fhinteger

The forecasting horizon. This represents the number of time steps into the future that the forecast should predict.

{"stackTrail":"components:schemas:SingleSeriesForecast:properties:y","oasType":"schema","type":"unknown","title":"Y","description":"The historical time series data provided as a dictionary. Each key is a timestamp (string format: YYYY-MM-DD) and the corresponding value is the observation at that time point. For example: {\"2021-01-01\": 0.1, \"2021-01-02\": 0.4}."}
xobject nullable

The exogenous variables provided as a dictionary. Each key is a timestamp (string format: YYYY-MM-DD) and the corresponding value is a list of exogenous variable values at that time point. For example: {"2021-01-01": [0.1], "2021-01-02": [0.4]}. This should also include forecasting horizon (fh) additional timestamps to calculate the future values.

clean_ex_firstboolean

A boolean flag that indicates whether the API should preprocess (clean) the exogenous signal before applying the large time model. If True, the exogenous signal is cleaned; if False, the exogenous variables are applied after the large time model.

finetune_stepsinteger

The number of tuning steps used to train the large time model on the data. Set this value to 0 for zero-shot inference, i.e., to make predictions without any further model tuning.

finetune_loss'default' | 'mae' | 'mse' | 'rmse' | 'mape' | 'smape'

The loss used to train the large time model on the data. Select from ['default', 'mae', 'mse', 'rmse', 'mape', 'smape']. It will only be used if finetune_steps larger than 0. Default is a robust loss function that is less sensitive to outliers.

Response

Successful Response

{"stackTrail":"paths:/forecast:post:responses:200:content:application/json:schema","oasType":"schema","type":"unknown"}

Changes