node-red-contrib-nixtla

Node-RED integration for nixtla

References

• Nixtla Documentation: https://docs.nixtla.io/docs

Project Links

• GitHub: https://github.com/Chiragasourabh/node-red-contrib-nixtla
• NPM: https://www.npmjs.com/package/node-red-contrib-nixtla
• NodeRed Project: https://flows.nodered.org/node/node-red-contrib-nixtla

Installation

Install via npm

npm install node-red-contrib-nixtla@latest

Usage

• Follow the steps mentioned here or click here to get your token.
• Drag and drop a nixtla node into node-Red Flow editor pane.
• Create a new nixtla configuration and paste the token.

Override values from previous node

The input values for the nixtla node can be overriden by passing respective properties from the previous node.

The topic for the nixtla node can be set as follows msg.topic

msg.topic = 'automl_forecast';
return msg;

The payload for the nixtla node can be set as follows msg.payload

msg = {}
payload = {};
payload.forecast_horizon = 1;
payload.timestamp = ["2022-05-10", "2022-05-11", "2022-05-12"];
payload.value = [0.5, 0.3, 0.1];
msg.payload = payload;
return msg;

Supported Topics:

"automl_forecast"
"automl_anomaly"
"forecast"
"neural_transfer"
"anomaly_detector"

Accepted properties in the payload for different topics:

"automl_forecast":

    "forecast_horizon" (int): Steps ahead you want to predict.
    "timestamp" (list): Each element of the list defines the timestamp of the time series.
    "value" (list): Time series values.

"automl_anomaly":
    
    "sensibility" (int): Confidence level for prediction intervals.
    "timestamp" (list): Each element of the list defines the timestamp of the time series.
    "value" (list): Time series values.

"forecast":

    "forecast_horizon" (int): Steps ahead you want to predict.
    "model" (str): Model name.
    "seasonality" (int): Seasonality
    "cv" (boolean): Whether to perform cross validation.
    "timestamp" (list): Each element of the list defines the timestamp of the time series.
    "value" (list): Time series values.

"neural_transfer":

    "forecast_horizon" (int): Steps ahead you want to predict.
    "model" (str): Model name.
    "max_steps" (int): K-shot learning steps.
    "timestamp" (list): Each element of the list defines the timestamp of the time series.
    "value" (list): Time series values.

"anomaly_detector":

    "forecast_horizon" (int): Steps ahead you want to predict.
    "sensibility" (int): Confidence level for prediction intervals.
    "seasonality" (int): Seasonality.
    "timestamp" (list): Each element of the list defines the timestamp of the time series.
    "value" (list): Time series values.

model

Statistical: 
    "arima" | "seasonal_exponential_smoothing" | "prophet" | "complex_es" | "ets"

Transfer Learning: 
    "nhits_m4_hourly" | "nhits_m4_hourly_tiny" | "nhits_m4_daily" | "nhits_m4_monthly" | "nhits_m4_yearly" | "nbeats_m4_hourly" | "nbeats_m4_daily" | "nbeats_m4_monthly" | "nbeats_m4_yearly"

Release

To release a new version of the package

• commit the changes
• create a tag with semantic version
• push the changes and tag

git tag -a v1.0.1 -m "release v1.0.1"
git push --follow-tags