XGBoost Parameters

Before running XGBoost, we must set three types of parameters: general parameters, booster parameters and task parameters.

• General parameters relate to which booster we are using to do boosting, commonly tree or linear model

• Booster parameters depend on which booster you have chosen

• Learning task parameters decide on the learning scenario. For example, regression tasks may use different parameters with ranking tasks.

• Command line parameters relate to behavior of CLI version of XGBoost.

Note

Parameters in R package

In R-package, you can use . (dot) to replace underscore in the parameters, for example, you can use max.depth to indicate max_depth. The underscore parameters are also valid in R.

• Global Configuration

• General Parameters

  • Parameters for Tree Booster

  • Additional parameters for hist and gpu_hist tree method

  • Additional parameters for gpu_hist tree method

  • Additional parameters for Dart Booster (booster=dart)

  • Parameters for Linear Booster (booster=gblinear)

  • Parameters for Tweedie Regression (objective=reg:tweedie)

• Learning Task Parameters

• Command Line Parameters

Global Configuration

The following parameters can be set in the global scope, using xgb.config_context() (Python) or xgb.set.config() (R).

• verbosity: Verbosity of printing messages. Valid values of 0 (silent), 1 (warning), 2 (info), and 3 (debug).

General Parameters

• booster [default= gbtree ]

  • Which booster to use. Can be gbtree, gblinear or dart; gbtree and dart use tree based models while gblinear uses linear functions.

• verbosity [default=1]

  • Verbosity of printing messages. Valid values are 0 (silent), 1 (warning), 2 (info), 3 (debug). Sometimes XGBoost tries to change configurations based on heuristics, which is displayed as warning message. If there’s unexpected behaviour, please try to increase value of verbosity.

• validate_parameters [default to false, except for Python, R and CLI interface]

  • When set to True, XGBoost will perform validation of input parameters to check whether a parameter is used or not. The feature is still experimental. It’s expected to have some false positives.

• nthread [default to maximum number of threads available if not set]

  • Number of parallel threads used to run XGBoost. When choosing it, please keep thread contention and hyperthreading in mind.

• disable_default_eval_metric [default=``false``]

  • Flag to disable default metric. Set to 1 or true to disable.

• num_pbuffer [set automatically by XGBoost, no need to be set by user]

  • Size of prediction buffer, normally set to number of training instances. The buffers are used to save the prediction results of last boosting step.

• num_feature [set automatically by XGBoost, no need to be set by user]

  • Feature dimension used in boosting, set to maximum dimension of the feature

Parameters for Tree Booster

• eta [default=0.3, alias: learning_rate]

  • Step size shrinkage used in update to prevents overfitting. After each boosting step, we can directly get the weights of new features, and eta shrinks the feature weights to make the boosting process more conservative.

  • range: [0,1]

• gamma [default=0, alias: min_split_loss]

  • Minimum loss reduction required to make a further partition on a leaf node of the tree. The larger gamma is, the more conservative the algorithm will be.

  • range: [0,∞]

• max_depth [default=6]

  • Maximum depth of a tree. Increasing this value will make the model more complex and more likely to overfit. 0 is only accepted in lossguided growing policy when tree_method is set as hist or gpu_hist and it indicates no limit on depth. Beware that XGBoost aggressively consumes memory when training a deep tree.

  • range: [0,∞] (0 is only accepted in lossguided growing policy when tree_method is set as hist or gpu_hist)

• min_child_weight [default=1]

  • Minimum sum of instance weight (hessian) needed in a child. If the tree partition step results in a leaf node with the sum of instance weight less than min_child_weight, then the building process will give up further partitioning. In linear regression task, this simply corresponds to minimum number of instances needed to be in each node. The larger min_child_weight is, the more conservative the algorithm will be.

  • range: [0,∞]

• max_delta_step [default=0]

  • Maximum delta step we allow each leaf output to be. If the value is set to 0, it means there is no constraint. If it is set to a positive value, it can help making the update step more conservative. Usually this parameter is not needed, but it might help in logistic regression when class is extremely imbalanced. Set it to value of 1-10 might help control the update.

  • range: [0,∞]

• subsample [default=1]

  • Subsample ratio of the training instances. Setting it to 0.5 means that XGBoost would randomly sample half of the training data prior to growing trees. and this will prevent overfitting. Subsampling will occur once in every boosting iteration.

  • range: (0,1]

• sampling_method [default= uniform]

  • The method to use to sample the training instances.

  • uniform: each training instance has an equal probability of being selected. Typically set subsample >= 0.5 for good results.

  • gradient_based: the selection probability for each training instance is proportional to the regularized absolute value of gradients (more specifically, g2+λh2−−−−−−−√$\sqrt{g^2+\lambda h^2}$). subsample may be set to as low as 0.1 without loss of model accuracy. Note that this sampling method is only supported when tree_method is set to gpu_hist; other tree methods only support uniform sampling.

• colsample_bytree, colsample_bylevel, colsample_bynode [default=1]

  • This is a family of parameters for subsampling of columns.

  • All colsample_by* parameters have a range of (0, 1], the default value of 1, and specify the fraction of columns to be subsampled.

  • colsample_bytree is the subsample ratio of columns when constructing each tree. Subsampling occurs once for every tree constructed.

  • colsample_bylevel is the subsample ratio of columns for each level. Subsampling occurs once for every new depth level reached in a tree. Columns are subsampled from the set of columns chosen for the current tree.

  • colsample_bynode is the subsample ratio of columns for each node (split). Subsampling occurs once every time a new split is evaluated. Columns are subsampled from the set of columns chosen for the current level.

  • colsample_by* parameters work cumulatively. For instance, the combination {'colsample_bytree':0.5, 'colsample_bylevel':0.5, 'colsample_bynode':0.5} with 64 features will leave 8 features to choose from at each split.

    On Python interface, when using hist, gpu_hist or exact tree method, one can set the feature_weights for DMatrix to define the probability of each feature being selected when using column sampling. There’s a similar parameter for fit method in sklearn interface.

• lambda [default=1, alias: reg_lambda]

  • L2 regularization term on weights. Increasing this value will make model more conservative.

• alpha [default=0, alias: reg_alpha]

  • L1 regularization term on weights. Increasing this value will make model more conservative.

• tree_method string [default= auto]

  • The tree construction algorithm used in XGBoost. See description in the reference paper and XGBoost Tree methods.

  • XGBoost supports approx, hist and gpu_hist for distributed training. Experimental support for external memory is available for approx and gpu_hist.

  • Choices: auto, exact, approx, hist, gpu_hist, this is a combination of commonly used updaters. For other updaters like refresh, set the parameter updater directly.

    • auto: Use heuristic to choose the fastest method.

      • For small dataset, exact greedy (exact) will be used.

      • For larger dataset, approximate algorithm (approx) will be chosen. It’s recommended to try hist and gpu_hist for higher performance with large dataset. (gpu_hist)has support for external memory.

      • Because old behavior is always use exact greedy in single machine, user will get a message when approximate algorithm is chosen to notify this choice.

    • exact: Exact greedy algorithm. Enumerates all split candidates.

    • approx: Approximate greedy algorithm using quantile sketch and gradient histogram.

    • hist: Faster histogram optimized approximate greedy algorithm.

    • gpu_hist: GPU implementation of hist algorithm.

• sketch_eps [default=0.03]

  • Only used for tree_method=approx.

  • This roughly translates into O(1 / sketch_eps) number of bins. Compared to directly select number of bins, this comes with theoretical guarantee with sketch accuracy.

  • Usually user does not have to tune this. But consider setting to a lower number for more accurate enumeration of split candidates.

  • range: (0, 1)

• scale_pos_weight [default=1]

  • Control the balance of positive and negative weights, useful for unbalanced classes. A typical value to consider: sum(negative instances) / sum(positive instances). See Parameters Tuning for more discussion. Also, see Higgs Kaggle competition demo for examples: R, py1, py2, py3.

• updater [default= grow_colmaker,prune]

  • A comma separated string defining the sequence of tree updaters to run, providing a modular way to construct and to modify the trees. This is an advanced parameter that is usually set automatically, depending on some other parameters. However, it could be also set explicitly by a user. The following updaters exist:

    • grow_colmaker: non-distributed column-based construction of trees.

    • grow_histmaker: distributed tree construction with row-based data splitting based on global proposal of histogram counting.

    • grow_local_histmaker: based on local histogram counting.

    • grow_quantile_histmaker: Grow tree using quantized histogram.

    • grow_gpu_hist: Grow tree with GPU.

    • sync: synchronizes trees in all distributed nodes.

    • refresh: refreshes tree’s statistics and/or leaf values based on the current data. Note that no random subsampling of data rows is performed.

    • prune: prunes the splits where loss < min_split_loss (or gamma).

  • In a distributed setting, the implicit updater sequence value would be adjusted to grow_histmaker,prune by default, and you can set tree_method as hist to use grow_histmaker.

• refresh_leaf [default=1]

  • This is a parameter of the refresh updater. When this flag is 1, tree leafs as well as tree nodes’ stats are updated. When it is 0, only node stats are updated.

• process_type [default= default]

  • A type of boosting process to run.

  • Choices: default, update

    • default: The normal boosting process which creates new trees.

    • update: Starts from an existing model and only updates its trees. In each boosting iteration, a tree from the initial model is taken, a specified sequence of updaters is run for that tree, and a modified tree is added to the new model. The new model would have either the same or smaller number of trees, depending on the number of boosting iteratons performed. Currently, the following built-in updaters could be meaningfully used with this process type: refresh, prune. With process_type=update, one cannot use updaters that create new trees.

• grow_policy [default= depthwise]

  • Controls a way new nodes are added to the tree.

  • Currently supported only if tree_method is set to hist.

  • Choices: depthwise, lossguide

    • depthwise: split at nodes closest to the root.

    • lossguide: split at nodes with highest loss change.

• max_leaves [default=0]

  • Maximum number of nodes to be added. Only relevant when grow_policy=lossguide is set.

• max_bin, [default=256]

  • Only used if tree_method is set to hist.

  • Maximum number of discrete bins to bucket continuous features.

  • Increasing this number improves the optimality of splits at the cost of higher computation time.

• predictor, [default=``auto``]

  • The type of predictor algorithm to use. Provides the same results but allows the use of GPU or CPU.

    • auto: Configure predictor based on heuristics.

    • cpu_predictor: Multicore CPU prediction algorithm.

    • gpu_predictor: Prediction using GPU. Used when tree_method is gpu_hist. When predictor is set to default value auto, the gpu_hist tree method is able to provide GPU based prediction without copying training data to GPU memory. If gpu_predictor is explicitly specified, then all data is copied into GPU, only recommended for performing prediction tasks.

• num_parallel_tree, [default=1] - Number of parallel trees constructed during each iteration. This option is used to support boosted random forest.

• monotone_constraints

  • Constraint of variable monotonicity. See tutorial for more information.

• interaction_constraints

  • Constraints for interaction representing permitted interactions. The constraints must be specified in the form of a nest list, e.g. [[0, 1], [2, 3, 4]], where each inner list is a group of indices of features that are allowed to interact with each other. See tutorial for more information

Additional parameters for hist and gpu_hist tree method

• single_precision_histogram, [default=``false``]

  • Use single precision to build histograms instead of double precision.

Additional parameters for gpu_hist tree method

• deterministic_histogram, [default=``true``]

  • Build histogram on GPU deterministically. Histogram building is not deterministic due to the non-associative aspect of floating point summation. We employ a pre-rounding routine to mitigate the issue, which may lead to slightly lower accuracy. Set to false to disable it.

Additional parameters for Dart Booster (booster=dart)

Note

Using predict() with DART booster

If the booster object is DART type, predict() will perform dropouts, i.e. only some of the trees will be evaluated. This will produce incorrect results if data is not the training data. To obtain correct results on test sets, set ntree_limit to a nonzero value, e.g.

preds = bst.predict(dtest, ntree_limit=num_round)

• sample_type [default= uniform]

  • Type of sampling algorithm.

    • uniform: dropped trees are selected uniformly.

    • weighted: dropped trees are selected in proportion to weight.

• normalize_type [default= tree]

  • Type of normalization algorithm.

    • tree: new trees have the same weight of each of dropped trees.

      • Weight of new trees are 1 / (k + learning_rate).

      • Dropped trees are scaled by a factor of k / (k + learning_rate).

    • forest: new trees have the same weight of sum of dropped trees (forest).

      • Weight of new trees are 1 / (1 + learning_rate).

      • Dropped trees are scaled by a factor of 1 / (1 + learning_rate).

• rate_drop [default=0.0]

  • Dropout rate (a fraction of previous trees to drop during the dropout).

  • range: [0.0, 1.0]

• one_drop [default=0]

  • When this flag is enabled, at least one tree is always dropped during the dropout (allows Binomial-plus-one or epsilon-dropout from the original DART paper).

• skip_drop [default=0.0]

  • Probability of skipping the dropout procedure during a boosting iteration.

    • If a dropout is skipped, new trees are added in the same manner as gbtree.

    • Note that non-zero skip_drop has higher priority than rate_drop or one_drop.

  • range: [0.0, 1.0]

Parameters for Linear Booster (booster=gblinear)

• lambda [default=0, alias: reg_lambda]

  • L2 regularization term on weights. Increasing this value will make model more conservative. Normalised to number of training examples.

• alpha [default=0, alias: reg_alpha]

  • L1 regularization term on weights. Increasing this value will make model more conservative. Normalised to number of training examples.

• updater [default= shotgun]

  • Choice of algorithm to fit linear model

    • shotgun: Parallel coordinate descent algorithm based on shotgun algorithm. Uses ‘hogwild’ parallelism and therefore produces a nondeterministic solution on each run.

    • coord_descent: Ordinary coordinate descent algorithm. Also multithreaded but still produces a deterministic solution.

• feature_selector [default= cyclic]

  • Feature selection and ordering method

    • cyclic: Deterministic selection by cycling through features one at a time.

    • shuffle: Similar to cyclic but with random feature shuffling prior to each update.

    • random: A random (with replacement) coordinate selector.

    • greedy: Select coordinate with the greatest gradient magnitude. It has O(num_feature^2) complexity. It is fully deterministic. It allows restricting the selection to top_k features per group with the largest magnitude of univariate weight change, by setting the top_k parameter. Doing so would reduce the complexity to O(num_feature*top_k).

    • thrifty: Thrifty, approximately-greedy feature selector. Prior to cyclic updates, reorders features in descending magnitude of their univariate weight changes. This operation is multithreaded and is a linear complexity approximation of the quadratic greedy selection. It allows restricting the selection to top_k features per group with the largest magnitude of univariate weight change, by setting the top_k parameter.

• top_k [default=0]

  • The number of top features to select in greedy and thrifty feature selector. The value of 0 means using all the features.

Parameters for Tweedie Regression (objective=reg:tweedie)

• tweedie_variance_power [default=1.5]

  • Parameter that controls the variance of the Tweedie distribution var(y) ~ E(y)^tweedie_variance_power

  • range: (1,2)

  • Set closer to 2 to shift towards a gamma distribution

  • Set closer to 1 to shift towards a Poisson distribution.

Learning Task Parameters

Specify the learning task and the corresponding learning objective. The objective options are below:

• objective [default=reg:squarederror]

  • reg:squarederror: regression with squared loss.

  • reg:squaredlogerror: regression with squared log loss 12[log(pred+1)−log(label+1)]2$\frac{1}{2}[log(pred+1)-log(label+1){]}^2$. All input labels are required to be greater than -1. Also, see metric rmsle for possible issue with this objective.

  • reg:logistic: logistic regression

  • reg:pseudohubererror: regression with Pseudo Huber loss, a twice differentiable alternative to absolute loss.

  • binary:logistic: logistic regression for binary classification, output probability

  • binary:logitraw: logistic regression for binary classification, output score before logistic transformation

  • binary:hinge: hinge loss for binary classification. This makes predictions of 0 or 1, rather than producing probabilities.

  • count:poisson –poisson regression for count data, output mean of poisson distribution

    • max_delta_step is set to 0.7 by default in poisson regression (used to safeguard optimization)

  • survival:cox: Cox regression for right censored survival time data (negative values are considered right censored). Note that predictions are returned on the hazard ratio scale (i.e., as HR = exp(marginal_prediction) in the proportional hazard function h(t) = h0(t) * HR).

  • survival:aft: Accelerated failure time model for censored survival time data. See Survival Analysis with Accelerated Failure Time for details.

  • aft_loss_distribution: Probabilty Density Function used by survival:aft objective and aft-nloglik metric.

  • multi:softmax: set XGBoost to do multiclass classification using the softmax objective, you also need to set num_class(number of classes)

  • multi:softprob: same as softmax, but output a vector of ndata * nclass, which can be further reshaped to ndata * nclass matrix. The result contains predicted probability of each data point belonging to each class.

  • rank:pairwise: Use LambdaMART to perform pairwise ranking where the pairwise loss is minimized

  • rank:ndcg: Use LambdaMART to perform list-wise ranking where Normalized Discounted Cumulative Gain (NDCG) is maximized

  • rank:map: Use LambdaMART to perform list-wise ranking where Mean Average Precision (MAP) is maximized

  • reg:gamma: gamma regression with log-link. Output is a mean of gamma distribution. It might be useful, e.g., for modeling insurance claims severity, or for any outcome that might be gamma-distributed.

  • reg:tweedie: Tweedie regression with log-link. It might be useful, e.g., for modeling total loss in insurance, or for any outcome that might be Tweedie-distributed.

• base_score [default=0.5]

  • The initial prediction score of all instances, global bias

  • For sufficient number of iterations, changing this value will not have too much effect.

• eval_metric [default according to objective]

  • Evaluation metrics for validation data, a default metric will be assigned according to objective (rmse for regression, and logloss for classification, mean average precision for ranking)

  • User can add multiple evaluation metrics. Python users: remember to pass the metrics in as list of parameters pairs instead of map, so that latter eval_metric won’t override previous one

  • The choices are listed below:

    • rmse: root mean square error

    • rmsle: root mean square log error: 1N[log(pred+1)−log(label+1)]2−−−−−−−−−−−−−−−−−−−−−−−−−−√$\sqrt{\frac{1}{N}[log(pred+1)-log(label+1){]}^2}$. Default metric of reg:squaredlogerror objective. This metric reduces errors generated by outliers in dataset. But because log function is employed, rmsle might output nan when prediction value is less than -1. See reg:squaredlogerror for other requirements.

    • mae: mean absolute error

    • mape: mean absolute percentage error

    • mphe: mean Pseudo Huber error. Default metric of reg:pseudohubererror objective.

    • logloss: negative log-likelihood

    • error: Binary classification error rate. It is calculated as #(wrong cases)/#(all cases). For the predictions, the evaluation will regard the instances with prediction value larger than 0.5 as positive instances, and the others as negative instances.

    • error@t: a different than 0.5 binary classification threshold value could be specified by providing a numerical value through ‘t’.

    • merror: Multiclass classification error rate. It is calculated as #(wrong cases)/#(all cases).

    • mlogloss: Multiclass logloss.

    • auc: Area under the curve. Available for binary classification and learning-to-rank tasks.

    • aucpr: Area under the PR curve. Available for binary classification and learning-to-rank tasks.

    • ndcg: Normalized Discounted Cumulative Gain

    • map: Mean Average Precision

    • ndcg@n, map@n: ‘n’ can be assigned as an integer to cut off the top positions in the lists for evaluation.

    • ndcg-, map-, ndcg@n-, map@n-: In XGBoost, NDCG and MAP will evaluate the score of a list without any positive samples as 1. By adding “-” in the evaluation metric XGBoost will evaluate these score as 0 to be consistent under some conditions.

    • poisson-nloglik: negative log-likelihood for Poisson regression

    • gamma-nloglik: negative log-likelihood for gamma regression

    • cox-nloglik: negative partial log-likelihood for Cox proportional hazards regression

    • gamma-deviance: residual deviance for gamma regression

    • tweedie-nloglik: negative log-likelihood for Tweedie regression (at a specified value of the tweedie_variance_power parameter)

    • aft-nloglik: Negative log likelihood of Accelerated Failure Time model. See Survival Analysis with Accelerated Failure Time for details.

    • interval-regression-accuracy: Fraction of data points whose predicted labels fall in the interval-censored labels. Only applicable for interval-censored data. See Survival Analysis with Accelerated Failure Time for details.

• seed [default=0]

  • Random number seed. This parameter is ignored in R package, use set.seed() instead.

• seed_per_iteration [default=false]

  • Seed PRNG determnisticly via iterator number, this option will be switched on automatically on distributed mode.

Command Line Parameters

The following parameters are only used in the console version of XGBoost

• num_round

  • The number of rounds for boosting

• data

  • The path of training data

• test:data

  • The path of test data to do prediction

• save_period [default=0]

  • The period to save the model. Setting save_period=10 means that for every 10 rounds XGBoost will save the model. Setting it to 0 means not saving any model during the training.

• task [default= train] options: train, pred, eval, dump

  • train: training using data

  • pred: making prediction for test:data

  • eval: for evaluating statistics specified by eval[name]=filename

  • dump: for dump the learned model into text format

• model_in [default=NULL]

  • Path to input model, needed for test, eval, dump tasks. If it is specified in training, XGBoost will continue training from the input model.

• model_out [default=NULL]

  • Path to output model after training finishes. If not specified, XGBoost will output files with such names as 0003.model where 0003 is number of boosting rounds.

• model_dir [default= models/]

  • The output directory of the saved models during training

• fmap

  • Feature map, used for dumping model

• dump_format [default= text] options: text, json

  • Format of model dump file

• name_dump [default= dump.txt]

  • Name of model dump file

• name_pred [default= pred.txt]

  • Name of prediction file, used in pred mode

• pred_margin [default=0]

  • Predict margin instead of transformed probability