Parameter tuning
CatBoost provides a flexible interface for parameter tuning and can be configured to suit different tasks.
One-hot encoding
Sometimes when categorical features don't have a lot of values, one-hot encoding works well.
- N/A if training is performed on CPU in Pairwise scoring mode
- 255 if training is performed on GPU and the selected Ctr types require target data that is not available during the training
- 10 if training is performed in Ranking mode
- 2 if none of the conditions above is met
- N/A if training is performed on CPU in Pairwise scoring mode
- 255 if training is performed on GPU and the selected Ctr types require target data that is not available during the training
- 10 if training is performed in Ranking mode
- 2 if none of the conditions above is met
Command-line version parameters | Python parameters | R parameters | Description | Default value |
---|---|---|---|---|
--one-hot-max-size | one_hot_max_size | one_hot_max_size | Use one-hot encoding for all categorical features with a number of different values less than or equal to the given parameter value. Ctrs are not calculated for such features. | The default value depends on various conditions: |
Command-line version parameters | Python parameters | R parameters | Description | Default value |
---|---|---|---|---|
--one-hot-max-size | one_hot_max_size | one_hot_max_size | Use one-hot encoding for all categorical features with a number of different values less than or equal to the given parameter value. Ctrs are not calculated for such features. | The default value depends on various conditions: |
Number of trees
It is recommended to check that there is no obvious underfitting or overfitting before tuning any other parameters. In order to do this it is necessary to analyze the metric value on the validation dataset and select the appropriate number of iterations.
This can be done by setting the number of iterations to a large value, using the overfitting detector parameters and turning the use best model options on. In this case the resulting model contains only the first k best iterations, where k is the iteration with the best loss value on the validation dataset.
Also, the metric for choosing the best model may differ from the one used for optimizing the objective value. For example, it is possible to set the optimized function to Logloss and use the AUC function for the overfitting detector. To do so, use the evaluation metric parameter.
- Build the number of trees defined by the training parameters.
- Use the validation dataset to identify the iteration with the optimal value of the metric specified in --eval-metric (eval_metric).
- IncToDec
- Iter
- IncToDec — Ignore the overfitting detector when the threshold is reached and continue learning for the specified number of iterations after the iteration with the optimal metric value.
- Iter — Consider the model overfitted and stop training after the specified number of iterations since the iteration with the optimal metric value.
- Build the number of trees defined by the training parameters.
- Use the validation dataset to identify the iteration with the optimal value of the metric specified in --eval-metric (eval_metric).
- IncToDec
- Iter
- IncToDec — Ignore the overfitting detector when the threshold is reached and continue learning for the specified number of iterations after the iteration with the optimal metric value.
- Iter — Consider the model overfitted and stop training after the specified number of iterations since the iteration with the optimal metric value.
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
-i --iterations | iterations | iterations | The maximum number of trees that can be built when solving machine learning problems. When using other parameters that limit the number of iterations, the final number of trees may be less than the number specified in this parameter. |
--use-best-model | use_best_model | use_best_model | If this parameter is set, the number of trees that are saved in the resulting model is defined as follows: No trees are saved after this iteration. This option requires a validation dataset to be provided. |
--eval-metric | eval_metric | eval_metric | The metric used for overfitting detection (if enabled) and best model selection (if enabled). Some metrics support optional parameters (see the Objectives and metrics section for details on each metric). Format:
Examples:
|
Overfitting detection settings | |||
--od-type | od_type | od_type | The type of the overfitting detector to use. Possible values: |
--od-pval | od_pval | od_pval | The threshold for the IncToDec overfitting detector type. The training is stopped when the specified value is reached. Requires that a validation dataset was input. For best results, it is recommended to set a value in the range . The larger the value, the earlier overfitting is detected. Restriction. Do not use this parameter with the Iter overfitting detector type. |
--od-wait | od_wait | od_wait | The number of iterations to continue the training after the iteration with the optimal metric value. The purpose of this parameter differs depending on the selected overfitting detector type: |
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
-i --iterations | iterations | iterations | The maximum number of trees that can be built when solving machine learning problems. When using other parameters that limit the number of iterations, the final number of trees may be less than the number specified in this parameter. |
--use-best-model | use_best_model | use_best_model | If this parameter is set, the number of trees that are saved in the resulting model is defined as follows: No trees are saved after this iteration. This option requires a validation dataset to be provided. |
--eval-metric | eval_metric | eval_metric | The metric used for overfitting detection (if enabled) and best model selection (if enabled). Some metrics support optional parameters (see the Objectives and metrics section for details on each metric). Format:
Examples:
|
Overfitting detection settings | |||
--od-type | od_type | od_type | The type of the overfitting detector to use. Possible values: |
--od-pval | od_pval | od_pval | The threshold for the IncToDec overfitting detector type. The training is stopped when the specified value is reached. Requires that a validation dataset was input. For best results, it is recommended to set a value in the range . The larger the value, the earlier overfitting is detected. Restriction. Do not use this parameter with the Iter overfitting detector type. |
--od-wait | od_wait | od_wait | The number of iterations to continue the training after the iteration with the optimal metric value. The purpose of this parameter differs depending on the selected overfitting detector type: |
Learning rate
This setting is used for reducing the gradient step. It affects the overall time of training: the smaller the value, the more iterations are required for training. Choose the value based on the performance expectations.
By default, the learning rate is defined automatically based on the dataset properties and the number of iterations. The automatically defined value should be close to the optimal one.
- There is no overfitting on the last iterations of training (the training does not converge) — increase the learning rate.
- Overfitting is detected — decrease the learning rate.
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
-w --learning-rate | learning_rate | learning_rate | The learning rate. Used for reducing the gradient step. |
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
-w --learning-rate | learning_rate | learning_rate | The learning rate. Used for reducing the gradient step. |
Tree depth
In most cases, the optimal depth ranges from 4 to 10. Values in the range from 6 to 10 are recommended.
The maximum depth of the trees is limited to 8 for pairwise modes (YetiRank, PairLogitPairwise and QueryCrossEntropy) when the training is performed on GPU.
CPU — Any integer up to 16.
GPU — Any integer up to 8 pairwise modes (YetiRank, PairLogitPairwise and QueryCrossEntropy) and up to 16 for all other loss functions.
CPU — Any integer up to 16.
GPU — Any integer up to 8 pairwise modes (YetiRank, PairLogitPairwise and QueryCrossEntropy) and up to 16 for all other loss functions.
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
-n --depth | depth | depth | Depth of the tree. The range of supported values depends on the processing unit type and the type of the selected loss function: |
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
-n --depth | depth | depth | Depth of the tree. The range of supported values depends on the processing unit type and the type of the selected loss function: |
L2 regularization
Try different values for the regularizer to find the best possible.
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
--l2-leaf-reg | l2_leaf_reg | l2_leaf_reg | Coefficient at the L2 regularization term of the cost function. Any positive value is allowed. |
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
--l2-leaf-reg | l2_leaf_reg | l2_leaf_reg | Coefficient at the L2 regularization term of the cost function. Any positive value is allowed. |
Random strength
Try setting different values for the random_strength
parameter.
- QueryCrossEntropy
- YetiRankPairwise
- PairLogitPairwise
- QueryCrossEntropy
- YetiRankPairwise
- PairLogitPairwise
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
--random-strength | random_strength | random_strength | The amount of randomness to use for scoring splits when the tree structure is selected. Use this parameter to avoid overfitting the model. The value of this parameter is used when selecting splits. On every iteration each possible split gets a score (for example, the score indicates how much adding this split will improve the loss function for the training dataset). The split with the highest score is selected. The scores have no randomness. A normally distributed random variable is added to the score of the feature. It has a zero mean and a variance that decreases during the training. The value of this parameter is the multiplier of the variance. Note. This parameter is not supported for the following loss functions: |
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
--random-strength | random_strength | random_strength | The amount of randomness to use for scoring splits when the tree structure is selected. Use this parameter to avoid overfitting the model. The value of this parameter is used when selecting splits. On every iteration each possible split gets a score (for example, the score indicates how much adding this split will improve the loss function for the training dataset). The split with the highest score is selected. The scores have no randomness. A normally distributed random variable is added to the score of the feature. It has a zero mean and a variance that decreases during the training. The value of this parameter is the multiplier of the variance. Note. This parameter is not supported for the following loss functions: |
Bagging temperature
Try setting different values for the bagging_temperature
parameter
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
--bagging-temperature | bagging_temperature | bagging_temperature | Defines the settings of the Bayesian bootstrap. It is used by default in classification and regression modes. Use the Bayesian bootstrap to assign random weights to objects. The weights are sampled from exponential distribution if the value of this parameter is set to “1”. All weights are equal to 1 if the value of this parameter is set to “0”. Possible values are in the range . The higher the value the more aggressive the bagging is. This parameter can be used if the selected bootstrap type is Bayesian. |
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
--bagging-temperature | bagging_temperature | bagging_temperature | Defines the settings of the Bayesian bootstrap. It is used by default in classification and regression modes. Use the Bayesian bootstrap to assign random weights to objects. The weights are sampled from exponential distribution if the value of this parameter is set to “1”. All weights are equal to 1 if the value of this parameter is set to “0”. Possible values are in the range . The higher the value the more aggressive the bagging is. This parameter can be used if the selected bootstrap type is Bayesian. |
Border count
The number of splits for numerical features.
The default value depends on the processing unit type and other parameters:
- CPU: 254
- GPU in PairLogitPairwise and YetiRankPairwise modes: 32
- GPU in all other modes: 128
The value of this parameter significantly impacts the speed of training on GPU. The smaller the value, the faster the training is performed (refer to the Number of splits for numerical features section for details).
128 splits are enough for many datasets. However, try to set the value of this parameter to 254 when training on GPU if the best possible quality is required.
The value of this parameter does not significantly impact the speed of training on CPU. Try to set it to 254 for the best possible quality.
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
-x --border-count | border_count Alias: max_bin | border_count | The number of splits for numerical features. Allowed values are integers from 1 to 65535 inclusively. Recommended values are up to 255. Larger values slow down the training. |
Command-line version parameters | Python parameters | R parameters | Description |
---|---|---|---|
-x --border-count | border_count Alias: max_bin |