lightgbm knime classifier

This KNIME node performs classification using lightgbm.LGBMClassifier. The classfier is compatible with sklearn API. All that the user has to do is to configure KNIME to use Python environment. The node when executed the first time creates a conda environment and installs lightgbm and related packages. The predictor node is 'lightgbm classifier predict' that you may download separately from KNIME hub. Complete details of parameters used in the widget are available here: https://lightgbm.readthedocs.io/en/latest/pythonapi/lightgbm.LGBMClassifier.html#lightgbm-lgbmclassifier

Options

Select columns to consider for modeling. Include target column (in the right panel) also:: Select all columns (including target) that you want included while modeling.
Target: Target column
learning_rate (default: 0.1): Boosting learning rate.
subsample (default: 1.0): Subsample ratio of the training instance.
colsample_bytree (default: 1.0): Subsample ratio of columns when constructing each tree.
reg_alpha (default: 0.0): L1 regularization term on weights.
reg_lambda (default: 0.0): L2 regularization term on weights.
min_split_gain (default: 0.0): Minimum loss reduction required to make a further partition on a leaf node of the tree
num_leaves (default: 31): Maximum tree leaves for base learners.
max_depth (default: -1): Maximum tree depth for base learners, <=0 means no limit.
n_estimators (default: 100): Number of boosted trees to fit.
subsample_for_bin (default: 200000): Number of samples for constructing bins.
min_child_samples (default: 20): Minimum number of data needed in a child (leaf).
boosting_type (default: gbdt): Defines the type of algorithm you want to run, to grow trees.
objective: Specify the learning task and the corresponding learning objective
is_unbalance (default: False or "-"): If data is unbalanced, set 'is_unbalance' to +, else -.

lightgbm knime classifier

Options

Input Ports

Output Ports

Nodes

Extensions

Links

Download