The purpose of this workflow is to showcase the ease of use of the H2O functionalities from within KNIME. As a real world usecase we chose the "Restaurant Visitor Forecasting" competition on Kaggle.com: https://www.kaggle.com/c/recruit-restaurant-visitor-forecasting
The workflow contains the following steps:
- Data preparation: Reading, cleaning, joining data and feature creation
- Creation of a local H2O context and transformation of a KNIME data table into an H2O frame
- Modeling of three different models including cross validation and parameter optimization
- Selection of the best model
- Deployment: Converting the H2O model into an H2O MOJO and doing the prediction for the Kaggle competition
Feel free to create some more features and try additional parameters in the optimization loop to improve your predictions.
For legal reasons we are not allowed to ship the dataset from Kaggle with our workflow. To get access to the data you have to sign in to Kaggle and accept the conditions of participation for the competetion. Afterwards you can download the data, save it in the data folder of this KNIME project and run the workflow.
External resources
Used extensions & nodes
Created with KNIME Analytics Platform version 4.3.0
-
KNIME Active Learning
KNIME AG, Zurich, Switzerland
Version 4.1.0
-
KNIME Core
KNIME AG, Zurich, Switzerland
Version 4.1.0
-
KNIME Ensemble Learning Wrappers
KNIME AG, Zurich, Switzerland
Version 4.1.0
-
KNIME H2O Machine Learning Integration
KNIME AG, Zurich, Switzerland
Versions 4.1.0, 4.3.0
-
KNIME H2O Machine Learning Integration - MOJO Extension
KNIME AG, Zurich, Switzerland
Version 4.1.0
-
KNIME Math Expression (JEP)
KNIME AG, Zurich, Switzerland
Version 4.1.0
-
KNIME Optimization extension
KNIME AG, Zurich, Switzerland
Version 4.3.0
Legal
By downloading the workflow, you agree to our terms and conditions.
License (CC-BY-4.0)Discussion
Discussions are currently not available, please try again later.