QuantixAI API Documentation

Upload a dataset as a JSON payload. The uploaded dataset will be stored and made available for analysis and forecasting. If a dataset with the same name exists, set overwrite accordingly.

Request Body

Example Request

curl -X POST https://data.quantix-ai.eu/api/v1/data/upload \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Sales Data Q4",
    "description": "Quarterly sales report",
    "preserve": true,
    "data": {
      "date": ["2024-01-01", "2024-01-02", "2024-01-03"],
      "sales": [100, 150, 200],
      "region": ["North", "South", "East"]
    }
  }'

Success Response

{
  "success": true,
  "file_id": "abc123def456",
  "message": "Data uploaded successfully"
}

Error Responses

Runs a causal impact analysis on your time series data. Provide the response variable and optional covariates, along with pre‑ and post‑intervention periods and seasonality settings.

Request Body (JSON)

Data Format

The data object should include the response variable under the key y and optional control series under keys like x1, x2, etc. Each key maps to an array of numeric values of equal length.

Example Request

POST /causalimpact
Host: causalimpact.quantix-ai.eu
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "data": {
    "y":  [100, 105, 110, 115, 108, 112, 120, 125, 130, 135],
    "x1": [50,  52,  51,  53,  55,  54,  56,  58,  57,  59]
  },
  "preperiod":  [1, 6],
  "postperiod": [7, 10],
  "CI": 0.95,
  "nseasons": 7
}

Success Response

The response contains a summary of the average and cumulative effects and a detailed series of predictions and effects.

{
  "summary": {
    "Metric": ["Actual", "Pred", "Pred.lower", "Pred.upper", ...],
    "Average": [150.5, 142.3, 138.1, 146.5, ...],
    "Cumulative": [1505, 1423, 1381, 1465, ...]
  },
  "series": {
    "response":         [100, 105, 110, 115, ...],
    "cum.response":     [100, 205, 315, ...],
    "point.pred":       [99.8, 105.2, 109.7, ...],
    "point.pred.lower": [95.1, 100.3, 104.8, ...],
    "point.pred.upper": [104.5, 110.1, 114.6, ...],
    "point.effect":     [0.2, -0.2, 0.3, ...],
    "cum.effect":       [0.2, 0.0, 0.3, ...]
    // Additional fields omitted for brevity
  }
}

Response Fields

Basic health check endpoint for the Causal Impact API service.

Authentication

This endpoint does not require authentication.

Example Response

{
  "status": "OK",
  "timestamp": "2024-12-23 15:30:45",
  "database": "connected"
}

Queue a training job for a machine‑learning forecasting model. Provide your time series data, the desired forecast horizon, data frequency and model metadata. Training runs asynchronously: the API returns a job_id immediately and automatically generates a model_id which you will use for forecasting and refitting. Poll the queue endpoints to track job progress.

Request Body

Response

{
  "job_id": "job_abc123",
  "message": "Training job queued successfully. Job ID: job_abc123",
  "model_id": "job_abc123",
  "status": "queued",
  "user_id": null,
  "metadata": {
    "status": "queued",
    "job_id": "job_abc123",
    "queue_position": 0,
    "estimated_wait_seconds": 30,
    "check_status_url": "/ml/job/job_abc123"
  }
}

Generate forecasts synchronously using a trained machine‑learning model. This endpoint blocks until completion and is intended for development or single‑user scenarios. Production systems should use /ml/forecast-async to avoid HTTP timeouts.

Request Body

Response

{
  "predictions": {
    "series1": [110.5, 115.2],
    "series2": [205.1, 210.3]
  },
  "model_id": "ml_abc123",
  "prediction_intervals": {
    "series1": {
      "lo-95": [100.2, 104.8],
      "hi-95": [120.8, 125.6]
    }
  },
  "fit_data": {
    "series1": {
      "dates": ["2023-06-01", "2023-07-01", "2023-08-01", "2023-09-01", "2023-10-01"],
      "values": [90, 92, 95, 98, 100]
    },
    "series2": {
      "dates": ["2023-06-01", "2023-07-01", "2023-08-01", "2023-09-01", "2023-10-01"],
      "values": [180, 185, 190, 195, 200]
    }
  }
}

Generate forecasts asynchronously using a trained ML model. This endpoint returns immediately with a job_id and is recommended for production applications.

Benefits

• ✅ No HTTP timeouts – clients can poll for results.
• ✅ Fair FIFO queueing for concurrent users.
• ✅ Progress updates available via polling.
• ✅ Scales to multiple concurrent users.

Request Body

Same parameters as /ml/forecast.

Initial Response

{
  "job_id": "abc123-def456-789",
  "status": "queued",
  "message": "Forecast job queued successfully",
  "metadata": {
    "queue_position": 2,
    "estimated_wait_seconds": 300,
    "estimated_wait_minutes": 5.0,
    "check_status_url": "/queue/job/abc123-def456-789"
  }
}

Refit an existing model with new data. This unified endpoint works for both ML and statistical models. It automatically detects the model type and updates model weights while preserving the original configuration (frequency, horizon, exogenous settings, static features and reconciliation structure). Use it to keep your model up‑to‑date without a full retraining.

Request Body

Response

{
  "status": "queued",
  "job_id": "refit_job_abc123",
  "message": "ML refit job queued successfully",
  "model_id": "your_model_id",
  "model_type": "ML"
}

Generate multiple forecast scenarios asynchronously using different exogenous assumptions. Each row in the data array must specify a scenario label and include values for all exogenous variables used during training. This endpoint returns a job_id; poll the queue to retrieve results.

Request Body

Initial Response

{
  "job_id": "scenario_abc123-def456",
  "status": "queued",
  "message": "Scenario forecast job queued successfully"
}

Generate forecast scenarios synchronously. Accepts the same payload as /ml/scenario-forecast-async but returns the results directly. Intended for testing or single‑user environments.

Train classical statistical models on your time series. The API automatically trains multiple model types and selects the best model per series based on cross‑validation performance. Statistical training runs asynchronously and returns a job_id.

Request Body

Response

{
  "job_id": "job_stats_123",
  "message": "Training job queued successfully. Job ID: job_stats_123",
  "model_id": "job_stats_123",
  "status": "queued",
  "user_id": null,
  "metadata": {
    "status": "queued",
    "job_id": "job_stats_123",
    "queue_position": 2,
    "estimated_wait_seconds": 45,
    "check_status_url": "/ml/job/job_stats_123"
  }
}

Generate forecasts synchronously using a trained statistical model. For non‑exogenous models you need only the model_id and horizon. For reconciled models you may optionally supply future data with exactly horizon rows per series.

Request Body

Response

{
    "predictions": {
        "series1": [110.5, 115.2],
        "series2": [205.1, 210.3]
    },
    "model_id": "stats_abc123",
    "prediction_intervals": {
        "series1": {
            "lo-95": [100.2, 104.8],
            "hi-95": [120.8, 125.6]
        }
    },
    "fit_data": {
        "series1": {
            "dates": ["2023-06-01", "2023-07-01", "2023-08-01", "2023-09-01", "2023-10-01"],
            "values": [90, 92, 95, 98, 100]
        },
        "series2": {
            "dates": ["2023-06-01", "2023-07-01", "2023-08-01", "2023-09-01", "2023-10-01"],
            "values": [180, 185, 190, 195, 200]
        }
    }
}

Generate forecasts asynchronously using a statistical model. This endpoint is recommended for production. The request body is the same as /stats/forecast.

Initial Response

{
  "job_id": "xyz789-abc123-456",
  "status": "queued",
  "message": "Forecast job queued successfully",
  "metadata": {
    "queue_position": 1,
    "estimated_wait_seconds": 180,
    "estimated_wait_minutes": 3.0,
    "check_status_url": "/queue/job/xyz789-abc123-456"
  }
}

Retrieve the status and result of an asynchronous job. Poll this endpoint every 2–5 seconds until the job’s status becomes finished or failed.

Response

{
  "job_id": "job_xyz789",
  "status": "completed",
  "created_at": "2024-01-15T10:30:00Z",
  "completed_at": "2024-01-15T10:35:00Z",
  "result": {
    "forecast": [...]
  },
  "error": null
}

Job Statuses

• queued – waiting in the queue
• processing – currently being processed
• completed – job completed successfully
• failed – job failed (see error field for details)

Retrieve statistics about all jobs in the queue, including counts of queued, processing, completed and failed jobs.

Response

{
  "total_jobs": 150,
  "queued": 5,
  "processing": 3,
  "completed": 140,
  "failed": 2
}

Cancel or delete a queued job. This action requires administrative privileges.

Health check endpoint for the queue service. Returns basic status information and does not require authentication.

Response

{
  "status": "healthy",
  "timestamp": "2024-01-15T10:30:00Z"
}

Returns the interactive documentation UI (for example, an OpenAPI/Swagger interface) for the Forecast API.