API Reference

Request body

{
  "inputs": {
    "age": 42,
    "city": "boston",
    "income": 61000
  }
}

Keys are your training file's column names (minus the target and any dropped columns). Send the raw values as they appeared in the file: numbers as numbers (numeric strings are fine), categories as the original text. Leave out anything you don't know — it's filled in and reported in warnings.

Response — 200 OK

{
  "prediction": "yes",          // number for regression, class label for classification
  "confidence": 0.83,           // 0–1 (null if unavailable)
  "job_id": "xxxxxxxx-...",
  "model_name": "Churn model",
  "all_probabilities": {        // multiclass models only
    "yes": 0.83, "no": 0.17
  },
  "warnings": [                 // only when inputs were filled in or ignored
    "'income' was missing; used the training median (58200)."
  ]
}

For regression, confidence is a heuristic based on how far the prediction is from the training mean, not a statistical interval.

Errors

Every error body is { "error": "description" }.

Python (requests)

import requests

r = requests.post(
    "https://cletus-service.onrender.com/predict/JOB_ID",
    json={"inputs": {"age": 42, "city": "boston", "income": 61000}},
    headers={"X-API-Key": "YOUR_API_KEY"},
    timeout=30,
)
r.raise_for_status()
result = r.json()
print(result["prediction"], result["confidence"], result.get("warnings", []))

JavaScript / TypeScript

const res = await fetch("https://cletus-service.onrender.com/predict/JOB_ID", {
  method: "POST",
  headers: { "Content-Type": "application/json", "X-API-Key": "YOUR_API_KEY" },
  body: JSON.stringify({ inputs: { age: 42, city: "boston", income: 61000 } }),
});
if (!res.ok) throw new Error((await res.json()).error);
const { prediction, confidence } = await res.json();

cURL

curl -sS -X POST "https://cletus-service.onrender.com/predict/JOB_ID" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{"inputs":{"age":42,"city":"boston","income":61000}}'

R (httr)

library(httr)
library(jsonlite)

res <- POST(
  "https://cletus-service.onrender.com/predict/JOB_ID",
  add_headers("X-API-Key" = "YOUR_API_KEY"),
  body = toJSON(list(inputs = list(age = 42, city = "boston", income = 61000)), auto_unbox = TRUE),
  content_type_json()
)
result <- content(res, "parsed")
cat(result$prediction, result$confidence, "\n")