Introduction to XGBoost Algorithm in Machine Learning

guest_blog 04 Jun, 2024
10 min read

XGBoost is a machine learning algorithm that belongs to the ensemble learning category, specifically the gradient boosting framework. It utilizes decision trees as base learners and employs regularization techniques to enhance model generalization. XGBoost is famous for its computational efficiency, offering efficient processing, insightful feature importance analysis, and seamless handling of missing values. It’s the go-to algorithm for a wide range of tasks, including regression, classification, and ranking. In this article, we will give you an overview of XGBoost, along with a use-case!

We recommend going through the below article as well to fully understand the various terms and concepts mentioned in this article:

What is XGBoost in Machine Learning?

XGBoost, or eXtreme Gradient Boosting, is a machine learning algorithm under ensemble learning. It is trendy for supervised learning tasks, such as regression and classification. XGBoost builds a predictive model by combining the predictions of multiple individual models, often decision trees, in an iterative manner.

The algorithm works by sequentially adding weak learners to the ensemble, with each new learner focusing on correcting the errors made by the existing ones. It uses a gradient descent optimization technique to minimize a predefined loss function during training.

Key features of XGBoost Algorithm include its ability to handle complex relationships in data, regularization techniques to prevent overfitting and incorporation of parallel processing for efficient computation.

Why Ensemble Learning?

XGBoost is an ensemble learning method. Sometimes, it may not be sufficient to rely upon the results of just one machine learning model. Ensemble learning offers a systematic solution to combine the predictive power of multiple learners. The resultant is a single model which gives the aggregated output from several models.

The models that form the ensemble, also known as base learners, could be either from the same learning algorithm or different learning algorithms. Bagging and boosting are two widely used ensemble learners. Though these two techniques can be used with several statistical models, the most predominant usage has been with decision trees.

Let’s briefly discuss bagging before taking a more detailed look at the concept of gradient boosting.

Bagging

While decision trees are one of the most easily interpretable models, they exhibit highly variable behavior. Consider a single training dataset that we randomly split into two parts. Now, let’s use each part to train a decision tree in order to obtain two models.

When we fit both these models, they would yield different results. Decision trees are said to be associated with high variance due to this behavior. Bagging or boosting aggregation helps to reduce the variance in any learner. Several decision trees which are generated in parallel, form the base learners of bagging technique. Data sampled with replacement is fed to these learners for training. The final prediction is the averaged output from all the learners.

Boosting

In boosting, the trees are built sequentially such that each subsequent tree aims to reduce the errors of the previous tree. Each tree learns from its predecessors and updates the residual errors. Hence, the tree that grows next in the sequence will learn from an updated version of the residuals.

The base learners in boosting are weak learners in which the bias is high, and the predictive power is just a tad better than random guessing. Each of these weak learners contributes some vital information for prediction, enabling the boosting technique to produce a strong learner by effectively combining these weak learners. The final strong learner brings down both the bias and the variance.

In contrast to bagging techniques like Random Forest, in which trees are grown to their maximum extent, boosting makes use of trees with fewer splits. Such small trees, which are not very deep, are highly interpretable. Parameters like the number of trees or iterations, the rate at which the gradient boosting learns, and the depth of the tree, could be optimally selected through validation techniques like k-fold cross validation. Having a large number of trees might lead to overfitting. So, it is necessary to carefully choose the stopping criteria for boosting.

Gradient Boosting Ensemble Technique

The gradient boosting ensemble technique consists of three simple steps:

  • An initial model F0 is defined to predict the target variable y. This model will be associated with a residual (y – F0)
  • A new model h1 is fit to the residuals from the previous step
  • Now, F0 and h1 are combined to give F1, the boosted version of F0. The mean squared error from F1 will be lower than that from F0:

To improve the performance of F1, we could model after the residuals of F1 and create a new model F2:

This can be done for ‘m’ iterations, until residuals have been minimized as much as possible:

Here, the additive learners do not disturb the functions created in the previous steps. Instead, they impart information of their own to bring down the errors.

Demonstrating the Potential of Gradient Boosting

In this section, we will explore the power of gradient boosting, a machine learning technique, by building an ensemble model to predict salary based on years of experience. By utilizing regression trees and optimizing loss functions, we aim to showcase the significant reduction in error that gradient boosting can achieve.

Introduction to the Predictive Model

Consider the following data where the years of experience is predictor variable and salary (in thousand dollars) is the target. Using regression trees as base learners, we can create an ensemble model to predict the salary. For the sake of simplicity, we can choose square loss as our loss function and our objective would be to minimize the square error.

years and salary , XGBoost algorithm

Initializing the Model and Understanding Residuals

As the first step, the model should be initialized with a function F0(x). F0(x) should be a function which minimizes the loss function or MSE (mean squared error), in this case:

F0x.f0x,XGboost algorithm

Taking the first differential of the above equation with respect to γ, it is seen that the function minimizes at the mean i=1nyin. So, the boosting model could be initiated with:

F0(x) gives the predictions from the first stage of our model. Now, the residual error for each instance is (yi – F0(x)).

X,Y F0(x)

Building Additive Learners

We can use the residuals from F0(x) to create h1(x). h1(x) will be a regression tree which will try and reduce the residuals from the previous step. The output of h1(x) won’t be a prediction of y; instead, it will help in predicting the successive function F1(x) which will bring down the residuals.

h1(x), xgboost algorithm

The additive model h1(x) computes the mean of the residuals (y – F0) at each leaf of the tree. The boosted function F1(x) is obtained by summing F0(x) and h1(x). This way h1(x) learns from the residuals of F0(x) and suppresses it in F1(x).

F0x,f1x

This can be repeated for 2 more iterations to compute h2(x) and h3(x). Each of these additive learners, hm(x), will make use of the residuals from the preceding function, Fm-1(x).

xgboost h2x

The MSEs for F0(x), F1(x) and F2(x) are 875, 692 and 540. It’s amazing how these simple weak learners can bring about a huge reduction in error!

Observing the Reduction in Error

Note that each learner, hm(x), is trained on the residuals. All the additive learners in boosting are modeled after the residual errors at each step. Intuitively, it could be observed that the boosting learners make use of the patterns in residual errors. At the stage where maximum accuracy is reached by boosting, the residuals appear to be randomly distributed without any pattern.

salry chart, xgboost algorithm
year of experience, xgboost
years of experience 2
Residuals,XGboost
salary ,years of experience
Plots of fn and hn
Plots of Fn and hn

Using Gradient d=Descent for Optimizing the Loss Function

In the case discussed above, MSE was the loss function. The mean minimized the error here. When MAE (mean absolute error) is the loss function, the median would be used as F₀(x) to initialize the model. A unit change in y would cause a unit change in MAE as well. Using scikit-learn, you can implement various models, including tree boosting algorithms and linear regression models to analyze the differences in loss functions and their impact on the model’s performance.

For MSE, the change observed would be roughly exponential. Instead of fitting hm(x) on the residuals, fitting it on the gradient of loss function, or the step along which loss occurs, would make this process generic and applicable across all loss functions.

Gradient descent helps us minimize any differentiable function. Earlier, the regression tree for hm(x) predicted the mean residual at each terminal node of the tree. In gradient boosting, the average gradient component would be computed.

For each node, there is a factor γ with which hm(x) is multiplied. This accounts for the difference in impact of each branch of the split. Gradient boosting helps in predicting the optimal gradient for the additive model, unlike classical gradient descent techniques which reduce error in the output at each iteration.

The following steps are involved in gradient boosting:

  • F0(x) – with which we initialize the boosting algorithm – is to be defined:
formula ,F0x
  • The gradient of the loss function is computed iteratively:
rim , formula,
  • Each hm(x) is fit on the gradient obtained at each step
  • The multiplicative factor γm for each terminal node is derived and the boosted model Fm(x) is defined:
multiplicatice factor ym

Unique Features of XGBoost Model

XGBoost is a popular implementation of gradient boosting. Let’s discuss some features or metrics of XGBoost that make it so interesting:

  • Regularization: XGBoost has an option to penalize complex models through both L1 and L2 regularization. Regularization helps in preventing overfitting
  • Handling sparse data: Missing values or data processing steps like one-hot encoding make data sparse. XGBoost Classifier incorporates a sparsity-aware split finding algorithm to handle different types of sparsity patterns in the data
  • Weighted quantile sketch: Most existing tree based algorithms can find the split points when the data points are of equal weights (using quantile sketch algorithm). However, they are not equipped to handle weighted data. XGBoost has a distributed weighted quantile sketch algorithm to effectively handle weighted data
  • Block structure for parallel learning: For faster computing, XGBoost Classifier can make use of multiple cores on the CPU. This is possible because of a block structure in its system design. Data is sorted and stored in in-memory units called blocks. Unlike other algorithms, this enables the data layout to be reused by subsequent iterations, instead of computing it again. This feature also serves useful for steps like split finding and column sub-sampling
  • Cache awareness: In XGBoost, Scala non-continuous memory access is required to get the gradient statistics by row index. Hence,tianqi chen XGBoost has been designed to make optimal use of hardware. This is true and done by allocating internal buffers in each thread, where the gradient statistics can be stored its Workflow. And these parallel tree make better XGboost algorithms with the help of julia and java lanuages.
  • Out-of-core computing: This feature optimizes the available disk space and maximizes its usage when handling huge datasets that do not fit into memory

Python Code for XGBoost

Here’s a live coding window to see how XGBoost works and play around with the code without leaving this article!

XGBoost Model Benefits and Attributes

  1. High accuracy: Xgboost Classifier is known for its accuracy and has been shown to outperform other machine learning algorithms in many predictive modeling tasks.
  2. Scalability: It is highly scalable and can handle large datasets with millions of rows and columns.
  3. Efficiency: It is designed to be computationally efficient and can quickly train models on large datasets.
  4. Flexibility: It supports a variety of data types and objectives, including regression, classification, and ranking problems.
  5. Regularization: It incorporates regularization techniques to avoid overfitting and improve generalization performance.
  6. Interpretability: It provides feature importance scores that can help users understand which features are most important for making predictions.
  7. Open-source: XGBoost Model is an open-source library that is widely used and supported by the data science community.

XGBoost vs Gradient Boosting

FeatureXGBoostGradient Boosting
DescriptionAdvanced implementation of gradient boostingEnsemble technique using weak learners
OptimizationRegularized objective functionError gradient minimization
EfficiencyHighly optimized, efficientComputationally intensive
Missing ValuesBuilt-in supportRequires preprocessing
RegularizationBuilt-in L1 and L2Requires external steps
Feature ImportanceBuilt-in measuresLimited, needs external calculation
InterpretabilityComplex, less interpretableMore interpretable models

Difference between XGBoost and Random Forest

FeatureXGBoostRandom Forest
DescriptionImproves mistakes from previous treesBuilds trees independently
Algorithm TypeBoostingBagging
Handling of Weak LearnersCorrects errors sequentiallyCombines predictions of independently built trees
RegularizationUses L1 and L2 regularization to prevent overfittingUsually doesn’t employ regularization techniques
PerformanceOften performs better on structured data but needs more tuningSimpler and less prone to overfitting

Conclusion

So that was all about the mathematics that power the popular XGBoost algorithm. If your basics are solid, this article must have been a breeze for you. It’s such a powerful algorithm and while there are other techniques that have spawned from it (like CATBoost), XGBoost Model remains a game changer in the machine learning community. We highly recommend you to take up this course to sharpen your skills in machine learning and learn all the state-of-the-art techniques used in the field with our Applied Machine Learning – Beginner to Professional course. Also, these Algorithm helps you for training data. and help you for learning rate that will help you for lightgbm the algorithms.

Frequently Asked Questions

Q1. Is XGBoost better than random forest?

A. The performance of XGBoost and random forest depends on the data and problem being solved. XGBoost tends to perform better on structured data, while random forest can be more effective on unstructured data.

Q2. What is XGBoost Python used for?

A. XGBoost Python is a Python package that enables building and training models using the XGBoost algorithm in Python. It includes many functions for tuning and optimizing model performance.

Q3. Is XGBoost a classifier or regression?

A. XGBoost is a versatile algorithm, applicable to both classification and regression tasks. It effectively manages various data types and can be tailored to meet specific requirements.

Q4. Is XGBoost a part of AI?

XGBoost is an AI tool. It’s a machine learning algorithm that helps AI systems make better predictions. You can also see it in github and kaggle.

Q5.What type of classifier is XGBoost?

XGBoost Classifier is a gradient boosting classifier. It’s a kind of ensemble classifier that improves by learning from past mistakes, often used for structured data tasks.

guest_blog 04 Jun, 2024

Frequently Asked Questions

Lorem ipsum dolor sit amet, consectetur adipiscing elit,

Responses From Readers

Clear

PRITESH S
PRITESH S 06 Sep, 2018

Nice explanation !

srinivas
srinivas 06 Sep, 2018

Hi. Nice article. Thanks for sharing. Couple of clarification 1. what's the formula for calculating the h1(X) 2. How did the split happen x23.

Anonymous
Anonymous 06 Sep, 2018

Do you have your app for iOS?

Kartheek Palepu
Kartheek Palepu 07 Sep, 2018

Great article. Can you just give a brief in the terms of regularization. What parameters get regularized? How the regularization happens in the case of multiple trees?

Sudhar
Sudhar 08 Sep, 2018

Very enlightening about the concept and interesting read. Can you brief me about loss functions?

sandoz
sandoz 10 Sep, 2018

Hi, I wish to propose to you a project using Xgboost with R. How can I contact you ? [email protected]

dave ro
dave ro 18 Sep, 2018

Thanks for sharing this great ariticle! Just have one clarification: h1 is calculated by some criterion(>23) on y-f0. In the resulted table, why there are some h1 with value 25.5 when y-f0 is negative (<23)?

Bert
Bert 15 Oct, 2018

Grate post! How this method treats outliers?

Murugan
Murugan 30 Oct, 2018

Its a great article. I have few clarifications: 1. How MSE is calculated. 2. Could you please explain in detail about the graphs.

Sudhir
Sudhir 31 Dec, 2018

Understandable explanation of boosting. However a small question For one of the question above you said, the split is made at x<=23 because it provides least SSE. In your excel did you went thru each of the x value as split to find the min SSE? While using XGBoost will it take its own the value of x (for min SSE) or we need to provide?

Lavjeet
Lavjeet 28 Jan, 2019

What is 1nyin?

ASHISH ANAND
ASHISH ANAND 16 Feb, 2019

Wow... You are awsome.. Thanks a lot for explaining in details...

Karthik
Karthik 17 Feb, 2019

A good interpretable explanation of the XGBoost algorithm! One thing I found confusing about the article is why we use a decision tree of depth 1 for making the first round of predictions and decision trees of depth 2 for making predictions on the residuals.

poorghorban
poorghorban 03 Mar, 2019

Thank you... It was very good article

ANONYMOUS
ANONYMOUS 11 Mar, 2019

Very clear. I am much grateful. There is one thing I could not wrap my head around. When devising h1(x) where the residuals are sorted according to x23, where does the criterion 23 come from?

Nick
Nick 05 May, 2022

Hello. May I clarify two points please: a) On the first step you calculate F0(x) as the argument that minimises the loss function. But isn't this the end? Once you have minimised the loss function why do you need to do anything else? b) On the 3rd step each hm(x) is fitted on the gradient obtained at each step. So you model the gradient? Would you please explain how exactly it is fitted? Thank you

mohit dev
mohit dev 11 Jul, 2022

Thank you for sharing quality content

mohit dev
mohit dev 11 Jul, 2022

Thank you for sharing quality content. keep sharing

Jake
Jake 11 Aug, 2023

Really clear explanation, thanks!