Icon

04 Advanced Machine Learning LAB

Advanced Machine Learning - Exercise (Solution)

This workflow shows a solution to a hands-on exercise in the L2-DS Introduction to KNIME Analytics Platform for Data Scientists - Advanced course

Task 2: Parameter Optimization Loop1. Start with a Parameter Optimization Loop Start node. Create a parameterfor the number of trees with start value=50, end value=150, andincrement=10. It is an integer.2. Overwrite the number of models setting in the Random Forest Learner(Regression) node with this parameter3. Transform the numeric scoring metrics into flow variables. Use the TableColumn to Variable node.4. End with a Parameter Optimization Loop End node. Use MAPE as theobjective value. Task 1: Random Forest (Regression)1. Execute the workflow. It reads and preprocesses the data about Airbnblistings in New York City, NY in 2019.2. Partition the data into a training set (70%) and a test set (30%). Applyrandom sampling.3. Train a Random Forest (Regression) model to predict the price column4. Apply the model to the test set5. Evaluate the model's performance with the Numeric Scorer node Predict priceR2 and error metricsControlnr of modelsMinimize MAPENode 101 InteractiveData Cleaning Random Forest Learner(Regression) Random Forest Predictor(Regression) Numeric Scorer Parameter OptimizationLoop Start ParameterOptimization Loop End Partitioning Replace 0 price byneighborhood average Table Columnto Variable CSV Reader Task 2: Parameter Optimization Loop1. Start with a Parameter Optimization Loop Start node. Create a parameterfor the number of trees with start value=50, end value=150, andincrement=10. It is an integer.2. Overwrite the number of models setting in the Random Forest Learner(Regression) node with this parameter3. Transform the numeric scoring metrics into flow variables. Use the TableColumn to Variable node.4. End with a Parameter Optimization Loop End node. Use MAPE as theobjective value. Task 1: Random Forest (Regression)1. Execute the workflow. It reads and preprocesses the data about Airbnblistings in New York City, NY in 2019.2. Partition the data into a training set (70%) and a test set (30%). Applyrandom sampling.3. Train a Random Forest (Regression) model to predict the price column4. Apply the model to the test set5. Evaluate the model's performance with the Numeric Scorer node Predict priceR2 and error metricsControlnr of modelsMinimize MAPENode 101InteractiveData Cleaning Random Forest Learner(Regression) Random Forest Predictor(Regression) Numeric Scorer Parameter OptimizationLoop Start ParameterOptimization Loop End Partitioning Replace 0 price byneighborhood average Table Columnto Variable CSV Reader

Nodes

Extensions

Links