ARIMA Learner

Trains an AutoRegressive Integrated Moving Average (ARIMA) model. ARIMA model captures temporal structures in time series data in the following components:
- AR: Relationship between the current observation and a number (p) of lagged observations
- I: Degree (d) of differencing required to make the time series stationary
- MA: Time series mean and the relationship between the current forecast error and a number (q) of lagged forecast errors

Additionally, coefficent statistics and residuals are provided as table outputs.

Model Summary metrics:
RMSE (Root Mean Square Error)
MAE (Mean Absolute Error)
MAPE (Mean Absolute Percentage Error)
*will be missing if zeroes in target
R2 (Coefficient of Determination)
Log Likelihood
AIC (Akaike Information Criterion)
BIC (Bayesian Information Criterion)

Note: This component requires a Python environment with StatsModels package installed. In this blog post we explain how to setup the KNIME Python extension:
https://www.knime.com/blog/setting-up-the-knime-python-extension-revisited-for-python-30-and-20

Python script is used due to performance reasons. KNIME Autoregressive integrated moving average (ARIMA) extension provides an alternative ARIMA Learner node:
https://kni.me/e/5_ZZ3nif8tLRjGji

Required extensions:
KNIME Python Integration
(https://hub.knime.com/knime/extensions/org.knime.features.python2/latest)
KNIME Quick Forms
(https://hub.knime.com/knime/extensions/org.knime.features.js.quickforms/latest)

Input ports

1. Type: Table
   Input data
   Table containing numeric target column to fit the ARIMA model.

Output ports

1. Type: Python
   ARIMA Model
   ARIMA model
2. Type: Table
   ARIMA Model Summary
   Table containing the coefficient statistics and the following evaluation metrics of the ARIMA model: RMSE MAE MAPE R2 Log Likelihood AIC BIC
3. Type: Table
   Residuals
   Table containing the residuals