Client

JSON goes to POST /predict. The Python Client posts Arrow to POST /predict/bytes. Same fields either way. The URL is a server you started.

Start a server

Point forecasts below use chronos_bolt. Quantile examples on the next pages use timesfm_2_5:

docker run --rm -p 8000:8000 sktime/tserve:hub chronos_bolt timesfm_2_5

Other tags and installers: Server. GET /models lists what this process loaded:

curl -s http://127.0.0.1:8000/models

Predict

Five days of sales, next three steps, chronos_bolt:

bash / zshPowerShell

curl -s http://127.0.0.1:8000/predict -H "Content-Type: application/json" -d '{
  "past": {
    "timestamp": ["2024-01-01", "2024-01-02", "2024-01-03", "2024-01-04", "2024-01-05"],
    "sales": [120, 135, 128, 142, 138]
  },
  "time": "timestamp",
  "target": ["sales"],
  "fh": 3,
  "model": "chronos_bolt"
}'

curl.exe -s http://127.0.0.1:8000/predict -H "Content-Type: application/json" -d '{"past":{"timestamp":["2024-01-01","2024-01-02","2024-01-03","2024-01-04","2024-01-05"],"sales":[120,135,128,142,138]},"time":"timestamp","target":["sales"],"fh":3,"model":"chronos_bolt"}'

The response is predictions, quantiles, model, and request_id. Fields and formats: Data specification.

In this section

• HTTP

  ---

  Send JSON to POST /predict from any language. Covers every route.

  HTTP

• Python

  ---

  Pass native tables to Client. It posts Arrow.

  Python

• Data specification

  ---

  Request fields, table formats, column inference, and response shape.

  Data specification