3. My first ANN

Workflow: My first Artificial Neural Network

This workflow reads the two datasets (training and test set) created in the workflow "1. Data Preparation" and further preprocesses the data. It then trains a Multilayer Perceptron (MLP), a feedforward artificial neural network, to predict the income (target class = "income"; >50k or <=50k) based on the remaining input features of the adult dataset.

Lastly, the trained ANN is evaluated based on the test set (ROC Curve node) and saved in the KBLData folder (Model Writer node).

Preprocessing

• Filter data: Keep only number columns & "income" (target column)

• Normalize data: Min-Max-normalization

Reading data

Training and test set created in workflow "1. Data Preparation".

Training ANN

Visualization

ROC curves and decision tree

Applying trained model

Save model

Save trained ANN to KBLdata folder