Training an H2O model from within KNIME
In this section, we will introduce the forecasting part of the application. We'll show how to train, optimize, evaluate, and deploy a stock price prediction model on H2O.
We selected the linear regression model as the stock price prediction model, which predicts the adjusted closing price based on the prices on the previous days. We can train the model as an H2O generalized linear model with Gaussian as the distribution family and Identity as the link function. This means we assume that the prices are approximately normally distributed and linearly related to the combination of the previous day's prices, as indicated by the following regression equation:
Formula 14.1
Where 
is the target variable, 
is the vector of the previous day's prices, 
is the vector of regression coefficients, and 
is the normally distributed residual term.
If we assumed a non-linear relationship between the target and predictor...
