Imbalanced dataset for benchmarking
=======================

The different algorithms of the `imbalanced-learn` toolbox are evaluated on a set of common dataset, which are more or less balanced. These benchmark have been proposed in [1]. The following section presents the main characteristics of this benchmark.

Characteristics
-------------------

|ID |Name |Repository & Target |Ratio |# samples| # features |
|:---:|:----------------------:|--------------------------------------|:------:|:-------------:|:--------------:|
|1 |Ecoli |UCI, target: imU |8.6:1 |336 |7 |
|2 |Optical Digits |UCI, target: 8 |9.1:1 |5,620 |64 |
|3 |SatImage |UCI, target: 4 |9.3:1 |6,435 |36 |
|4 |Pen Digits |UCI, target: 5 |9.4:1 |10,992 |16 |
|5 |Abalone |UCI, target: 7 |9.7:1 |4,177 |8 |
|6 |Sick Euthyroid |UCI, target: sick euthyroid |9.8:1 |3,163 |25 |
|7 |Spectrometer |UCI, target: >=44 |11:1 |531 |93 |
|8 |Car_Eval_34 |UCI, target: good, v good |12:1 |1,728 |6 |
|9 |ISOLET |UCI, target: A, B |12:1 |7,797 |617 |
|10 |US Crime |UCI, target: >0.65 |12:1 |1,994 |122 |
|11 |Yeast_ML8 |LIBSVM, target: 8 |13:1 |2,417 |103 |
|12 |Scene |LIBSVM, target: >one label |13:1 |2,407 |294 |
|13 |Libras Move |UCI, target: 1 |14:1 |360 |90 |
|14 |Thyroid Sick |UCI, target: sick |15:1 |3,772 |28 |
|15 |Coil_2000 |KDD, CoIL, target: minority |16:1 |9,822 |85 |
|16 |Arrhythmia |UCI, target: 06 |17:1 |452 |279 |
|17 |Solar Flare M0 |UCI, target: M->0 |19:1 |1,389 |10 |
|18 |OIL |UCI, target: minority |22:1 |937 |49 |
|19 |Car_Eval_4 |UCI, target: vgood |26:1 |1,728 |6 |
|20 |Wine Quality |UCI, wine, target: <=4 |26:1 |4,898 |11 |
|21 |Letter Img |UCI, target: Z |26:1 |20,000 |16 |
|22 |Yeast _ME2 |UCI, target: ME2 |28:1 |1,484 |8 |
|23 |Webpage |LIBSVM, w7a, target: minority|33:1 |49,749 |300 |
|24 |Ozone Level |UCI, ozone, data |34:1 |2,536 |72 |
|25 |Mammography |UCI, target: minority |42:1 |11,183 |6 |
|26 |Protein homo. |KDD CUP 2004, minority |111:1|145,751 |74 |
|27 |Abalone_19 |UCI, target: 19 |130:1|4,177 |8 |

References
----------
[1] Ding, Zejin, "Diversified Ensemble Classifiers for H
ighly Imbalanced Data Learning and their Application in Bioinformatics." Dissertation, Georgia State University, (2011).

[2] Blake, Catherine, and Christopher J. Merz. "UCI Repository of machine learning databases." (1998).

[3] Chang, Chih-Chung, and Chih-Jen Lin. "LIBSVM: a library for support vector machines." ACM Transactions on Intelligent Systems and Technology (TIST) 2.3 (2011): 27.

[4] Caruana, Rich, Thorsten Joachims, and Lars Backstrom. "KDD-Cup 2004: results and analysis." ACM SIGKDD Explorations Newsletter 6.2 (2004): 95-108.

Classification learning on non-stationary data may face dynamic changes from time to time. The major problem in it is the class imbalance and high cost of labeling instances despite drifts. Imbalance is due to lower number of samples in the minority class than the majority class. Imbalanced data results in the misclassification of data points.

In recent years, the challenge of imbalanced data has become increasingly prominent in machine learning, affecting the performance of classification algorithms. This study proposes a novel data-level oversampling method called Cluster-Based Reduced Noise SMOTE (CRN-SMOTE) to address this issue. CRN-SMOTE combines SMOTE for oversampling minority classes with a novel cluster-based noise reduction technique. In this cluster-based noise reduction approach, it is crucial that samples from each category form one or two clusters, a feature that conventional noise reduction methods do not achieve. The proposed method is evaluated on four imbalanced datasets (ILPD, QSAR, Blood, and Maternal Health Risk) using five metrics: Cohen’s kappa, Matthew’s correlation coefficient (MCC), F1-score, precision, and recall. Results demonstrate that CRN-SMOTE consistently outperformed the state-of-the-art Reduced Noise SMOTE (RN-SMOTE), SMOTE-Tomek Link, and SMOTE-ENN methods across all datasets, with particularly notable improvements observed in the QSAR and Maternal Health Risk datasets, indicating its effectiveness in enhancing imbalanced classification performance. Overall, the experimental findings indicate that CRN-SMOTE outperformed RN-SMOTE in 100% of the cases, achieving average improvements of 6.6% in Kappa, 4.01% in MCC, 1.87% in F1-score, 1.7% in precision, and 2.05% in recall, with setting SMOTE’s neighbors’ number to 5.

Dataset

This dataset was created by Jay Pradip Shah

Contents

This dataset file is used for the study of imbalanced data and contains 6 imbalanced datasets

Data sets supporting the results reported in the paper: Hellinger Distance Trees for Imbalanced Streams, R. J. Lyon, J.M. Brooke, J.D. Knowles, B.W Stappers, 22nd International Conference on Pattern Recognition (ICPR), p.1969 - 1974, 2014. DOI: 10.1109/ICPR.2014.344 Contained in this distribution are results of stream classifier perfromance on four different data sets. Also included are the test results from our attempt at reproducing the outcome of the paper, Learning Decision Trees for Un-balanced Data, D. A. Cieslak and N. V. Chawla, in Machine Learning and Knowledge Discovery in Databases (W. Daelemans, B. Goethals, and K. Morik, eds.), vol. 5211 of LNCS, pp. 241-256, 2008. The data sets used for these experiments include, MAGIC Gamma Telescope Data Set : https://archive.ics.uci.edu/ml/datasets/MAGIC+Gamma+TelescopeMiniBooNE particle identification Data Set : https://archive.ics.uci.edu/ml/datasets/MiniBooNE+particle+identificationSkin Segmentation Data Set : https://archive.ics.uci.edu/ml/datasets/Skin+SegmentationLetter Recognition Data Set : https://archive.ics.uci.edu/ml/datasets/Letter+RecognitionPen-Based Recognition of Handwritten Digits Data Set : https://archive.ics.uci.edu/ml/datasets/Pen-Based+Recognition+of+Handwritten+DigitsStatlog (Landsat Satellite) Data Set : https://archive.ics.uci.edu/ml/datasets/Statlog+(Landsat+Satellite)Statlog (Image Segmentation) Data Set : https://archive.ics.uci.edu/ml/datasets/Statlog+(Image+Segmentation) A further data set used is not publicly available at present. However we are in the process of releasing it for public use. Please get in touch if you'd like to use it.

A readme file accompanies the data describing it in more detail.

This dataset contains all the simulation results on the effect of ensemble models in dealing with data imbalance. The simulations are performed with sample size n=2000, number of variables p=200, and number of groups k=20 under six imbalanced scenarios. It shows the result of ensemble models with threshold from [0, 0.05, 0.1, ..., 0.95, 1.0], in terms of the overall AP/AR and discrete (continuous) specific AP/AR. This dataset serves as a reference for practitioners to find the appropriate ensemble threshold that fits their business needs the best.

Performance comparison of machine learning models across accuracy, AUC, MCC, and F1 score on GMSC dataset.

This record contains the underlying research data for the publication "High impact bug report identification with imbalanced learning strategies" and the full-text is available from: https://ink.library.smu.edu.sg/sis_research/3702In practice, some bugs have more impact than others and thus deserve more immediate attention. Due to tight schedule and limited human resources, developers may not have enough time to inspect all bugs. Thus, they often concentrate on bugs that are highly impactful. In the literature, high-impact bugs are used to refer to the bugs which appear at unexpected time or locations and bring more unexpected effects (i.e., surprise bugs), or break pre-existing functionalities and destroy the user experience (i.e., breakage bugs). Unfortunately, identifying high-impact bugs from thousands of bug reports in a bug tracking system is not an easy feat. Thus, an automated technique that can identify high-impact bug reports can help developers to be aware of them early, rectify them quickly, and minimize the damages they cause. Considering that only a small proportion of bugs are high-impact bugs, the identification of high-impact bug reports is a difficult task. In this paper, we propose an approach to identify high-impact bug reports by leveraging imbalanced learning strategies. We investigate the effectiveness of various variants, each of which combines one particular imbalanced learning strategy and one particular classification algorithm. In particular, we choose four widely used strategies for dealing with imbalanced data and four state-of-the-art text classification algorithms to conduct experiments on four datasets from four different open source projects. We mainly perform an analytical study on two types of high-impact bugs, i.e., surprise bugs and breakage bugs. The results show that different variants have different performances, and the best performing variants SMOTE (synthetic minority over-sampling technique) + KNN (K-nearest neighbours) for surprise bug identification and RUS (random under-sampling) + NB (naive Bayes) for breakage bug identification outperform the F1-scores of the two state-of-the-art approaches by Thung et al. and Garcia and Shihab.Supplementary code and data available from GitHub:

The classification models built on class imbalanced data sets tend to prioritize the accuracy of the majority class, and thus, the minority class generally has a higher misclassification rate. Different techniques are available to address the class imbalance in classification models and can be categorized as data-level, algorithm-level, and hybrid methods. But to the best of our knowledge, an in-depth analysis of the performance of these techniques against the class ratio is not available in the literature. We have addressed these shortcomings in this study and have performed a detailed analysis of the performance of four different techniques to address imbalanced class distribution using machine learning (ML) methods and AutoML tools. To carry out our study, we have selected four such techniques(a) threshold optimization using (i) GHOST and (ii) the area under the precision–recall curve (AUPR) curve, (b) internal balancing method of AutoML and class-weight of machine learning methods, and (c) data balancing using SMOTETomekand generated 27 data sets considering nine different class ratios (i.e., the ratio of the positive class and total samples) from three data sets that belong to the drug discovery and development field. We have employed random forest (RF) and support vector machine (SVM) as representatives of ML classifier and AutoGluon-Tabular (version 0.6.1) and H2O AutoML (version 3.40.0.4) as representatives of AutoML tools. The important findings of our studies are as follows: (i) there is no effect of threshold optimization on ranking metrics such as AUC and AUPR, but AUC and AUPR get affected by class-weighting and SMOTTomek; (ii) for ML methods RF and SVM, significant percentage improvement up to 375, 33.33, and 450 over all the data sets can be achieved, respectively, for F1 score, MCC, and balanced accuracy, which are suitable for performance evaluation of imbalanced data sets; (iii) for AutoML libraries AutoGluon-Tabular and H2O AutoML, significant percentage improvement up to 383.33, 37.25, and 533.33 over all the data sets can be achieved, respectively, for F1 score, MCC, and balanced accuracy; (iv) the general pattern of percentage improvement in balanced accuracy is that the percentage improvement increases when the class ratio is systematically decreased from 0.5 to 0.1; in the case of F1 score and MCC, maximum improvement is achieved at the class ratio of 0.3; (v) for both ML and AutoML with balancing, it is observed that any individual class-balancing technique does not outperform all other methods on a significantly higher number of data sets based on F1 score; (vi) the three external balancing techniques combined outperformed the internal balancing methods of the ML and AutoML; (vii) AutoML tools perform as good as the ML models and in some cases perform even better for handling imbalanced classification when applied with imbalance handling techniques. In summary, exploration of multiple data balancing techniques is recommended for classifying imbalanced data sets to achieve optimal performance as neither of the external techniques nor the internal techniques outperform others significantly. The results are specific to the ML methods and AutoML libraries used in this study, and for generalization, a study can be carried out considering a sizable number of ML methods and AutoML libraries.

Context

Insurance companies that sell life, health, and property and casualty insurance are using machine learning (ML) to drive improvements in customer service, fraud detection, and operational efficiency. The data provided by an Insurance company which is not excluded from other companies to getting advantage of ML. This company provides Health Insurance to its customers. We can build a model to predict whether the policyholders (customers) from past year will also be interested in Vehicle Insurance provided by the company.

An insurance policy is an arrangement by which a company undertakes to provide a guarantee of compensation for specified loss, damage, illness, or death in return for the payment of a specified premium. A premium is a sum of money that the customer needs to pay regularly to an insurance company for this guarantee.

For example, you may pay a premium of Rs. 5000 each year for a health insurance cover of Rs. 200,000/- so that if, God forbid, you fall ill and need to be hospitalized in that year, the insurance provider company will bear the cost of hospitalization etc. for up to Rs. 200,000. Now if you are wondering how can company bear such high hospitalization cost when it charges a premium of only Rs. 5000/-, that is where the concept of probabilities comes in picture. For example, like you, there may be 100 customers who would be paying a premium of Rs. 5000 every year, but only a few of them (say 2-3) would get hospitalized that year and not everyone. This way everyone shares the risk of everyone else.

Just like medical insurance, there is vehicle insurance where every year customer needs to pay a premium of certain amount to insurance provider company so that in case of unfortunate accident by the vehicle, the insurance provider company will provide a compensation (called ‘sum assured’) to the customer.

Content

Building a model to predict whether a customer would be interested in Vehicle Insurance is extremely helpful for the company because it can then accordingly plan its communication strategy to reach out to those customers and optimize its business model and revenue.

We have information about: - Demographics (gender, age, region code type), - Vehicles (Vehicle Age, Damage), - Policy (Premium, sourcing channel) etc.

Update: Test data target values has been added. To evaluate your models more precisely you can use: https://www.kaggle.com/arashnic/answer

#
#

Moreover the supplemental goal is to practice learning imbalanced data and verify how the results can help in real operational process. The Response feature (target) is highly imbalanced.

#

0: 319594 1: 62531 Name: Response, dtype: int64

#
Practicing some techniques like resampling is useful to verify impacts on validation results and confusion matrix. #
https://miro.medium.com/max/640/1*KxFmI15rxhvKRVl-febp-Q.png"> figure. Under-sampling: Tomek links # #

Starter Kernel(s)

• Quick EDA and LGB ,XGB
• Handling Imbalanced: Resampling the right way

Inspiration

Predict whether a customer would be interested in Vehicle Insurance

#
#

MORE DATASETs ...

This is a real-world industrial benchmark dataset from a major medical device manufacturer for the prediction of customer escalations. The dataset contains features derived from IoT (machine log) and enterprise data including labels for escalation from a fleet of thousands of customers of high-end medical devices.

The dataset accompanies the publication "System Design for a Data-driven and Explainable Customer Sentiment Monitor" (submitted). We provide an anonymized version of data collected over a period of two years.

The dataset should fuel the research and development of new machine learning algorithms to better cope with real-world data challenges including sparse and noisy labels, and concept drifts. Additional challenges is the optimal fusion of enterprise and log based features for the prediction task. Thereby, interpretability of designed prediction models should be ensured in order to have practical relevancy.

Supporting software

Kindly use the corresponding GitHub repository (https://github.com/annguy/customer-sentiment-monitor) to design and benchmark your algorithms.

Citation and Contact

If you use this dataset please cite the following publication:

@ARTICLE{9520354,
 author={Nguyen, An and Foerstel, Stefan and Kittler, Thomas and Kurzyukov, Andrey and Schwinn, Leo and Zanca, Dario and Hipp, Tobias and Jun, Sun Da and Schrapp, Michael and Rothgang, Eva and Eskofier, Bjoern},
 journal={IEEE Access}, 
 title={System Design for a Data-Driven and Explainable Customer Sentiment Monitor Using IoT and Enterprise Data}, 
 year={2021},
 volume={9},
 number={},
 pages={117140-117152},
 doi={10.1109/ACCESS.2021.3106791}}

If you would like to get in touch, please contact an.nguyen@fau.de.

Increase in the number of new chemicals synthesized in past decades has resulted in constant growth in the development and application of computational models for prediction of activity as well as safety profiles of the chemicals. Most of the time, such computational models and its application must deal with imbalanced chemical data. It is indeed a challenge to construct a classifier using imbalanced data set. In this study, we analyzed and validated the importance of different sampling methods over non-sampling method, to achieve a well-balanced sensitivity and specificity of a machine learning model trained on imbalanced chemical data. Additionally, this study has achieved an accuracy of 93.00%, an AUC of 0.94, F1 measure of 0.90, sensitivity of 96.00% and specificity of 91.00% using SMOTE sampling and Random Forest classifier for the prediction of Drug Induced Liver Injury (DILI). Our results suggest that, irrespective of data set used, sampling methods can have major influence on reducing the gap between sensitivity and specificity of a model. This study demonstrates the efficacy of different sampling methods for class imbalanced problem using binary chemical data sets.

Context

A stroke, also known as a cerebrovascular accident or CVA is when part of the brain loses its blood supply and the part of the body that the blood-deprived brain cells control stops working. This loss of blood supply can be ischemic because of lack of blood flow, or hemorrhagic because of bleeding into brain tissue. A stroke is a medical emergency because strokes can lead to death or permanent disability. There are opportunities to treat ischemic strokes but that treatment needs to be started in the first few hours after the signs of a stroke begin.

Content

The cerebral Stroke dataset consists of 12 features including the target column which is imbalanced.

Acknowledgements

Liu, Tianyu; Fan, Wenhui; Wu, Cheng (2019), “Data for A hybrid machine learning approach to cerebral stroke prediction based on imbalanced medical-datasets”, Mendeley Data, V1, doi: 10.17632/x8ygrw87jw.1 Dataset is sourced from here.

Clinical data analysis and forecasting have made substantial contributions to disease control, prevention and detection. However, such data usually suffer from highly imbalanced samples in class distributions. In this paper, we aim to formulate effective methods to rebalance binary imbalanced dataset, where the positive samples take up only the minority. We investigate two different meta-heuristic algorithms, particle swarm optimization and bat algorithm, and apply them to empower the effects of synthetic minority over-sampling technique (SMOTE) for pre-processing the datasets. One approach is to process the full dataset as a whole. The other is to split up the dataset and adaptively process it one segment at a time. The experimental results reported in this paper reveal that the performance improvements obtained by the former methods are not scalable to larger data scales. The latter methods, which we call Adaptive Swarm Balancing Algorithms, lead to significant efficiency and effectiveness improvements on large datasets while the first method is invalid. We also find it more consistent with the practice of the typical large imbalanced medical datasets. We further use the meta-heuristic algorithms to optimize two key parameters of SMOTE. The proposed methods lead to more credible performances of the classifier, and shortening the run time compared to brute-force method.

Credit Card Fraud Detection Dataset (European Cardholders, September 2013)

As a data contributor, I'm sharing this crucial dataset focused on the detection of fraudulent credit card transactions. Recognizing these illicit activities is paramount for protecting customers and the integrity of financial systems.

About the Dataset:

This dataset encompasses credit card transactions made by European cardholders during a two-day period in September 2013. It presents a real-world scenario with a significant class imbalance, where fraudulent transactions are considerably less frequent than legitimate ones. Out of a total of 284,807 transactions, only 492 are instances of fraud, representing a mere 0.172% of the entire dataset.

Content of the Data:

Due to confidentiality concerns, the majority of the input features in this dataset have undergone a Principal Component Analysis (PCA) transformation. This means the original meaning and context of features V1, V2, ..., V28 are not directly provided. However, these principal components capture the variance in the underlying transaction data.

The only features that have not been transformed by PCA are:

• Time: Numerical. Represents the number of seconds elapsed between each transaction and the very first transaction recorded in the dataset.
• Amount: Numerical. The transaction amount in Euros (€). This feature could be valuable for cost-sensitive learning approaches.

The target variable for this classification task is:

• Class: Integer. Takes the value 1 in the case of a fraudulent transaction and 0 otherwise.

Important Note on Evaluation:

Given the substantial class imbalance (far more legitimate transactions than fraudulent ones), traditional accuracy metrics based on the confusion matrix can be misleading. It is strongly recommended to evaluate models using the Area Under the Precision-Recall Curve (AUPRC), as this metric is more sensitive to the performance on the minority class (fraudulent transactions).

How to Use This Dataset:

1. Download the dataset file (likely in CSV format).
2. Load the data using libraries like Pandas.
3. Understand the class imbalance: Be aware that fraudulent transactions are rare.
4. Explore the features: Analyze the distributions of 'Time', 'Amount', and the PCA-transformed features (V1-V28).
5. Address the class imbalance: Consider using techniques like oversampling the minority class, undersampling the majority class, or using specialized algorithms designed for imbalanced datasets.
6. Build and train binary classification models to predict the 'Class' variable.
7. Evaluate your models using AUPRC to get a meaningful assessment of performance in detecting fraud.

Acknowledgements and Citation:

This dataset has been collected and analyzed through a research collaboration between Worldline and the Machine Learning Group (MLG) of ULB (Université Libre de Bruxelles).

When using this dataset in your research or projects, please cite the following works as appropriate:

• Andrea Dal Pozzolo, Olivier Caelen, Reid A. Johnson and Gianluca Bontempi. Calibrating Probability with Undersampling for Unbalanced Classification. In Symposium on Computational Intelligence and Data Mining (CIDM), IEEE, 2015.
• Dal Pozzolo, Andrea; Caelen, Olivier; Le Borgne, Yann-Ael; Waterschoot, Serge; Bontempi, Gianluca. Learned lessons in credit card fraud detection from a practitioner perspective, Expert systems with applications,41,10,4915-4928,2014, Pergamon.
• Dal Pozzolo, Andrea; Boracchi, Giacomo; Caelen, Olivier; Alippi, Cesare; Bontempi, Gianluca. Credit card fraud detection: a realistic modeling and a novel learning strategy, IEEE transactions on neural networks and learning systems,29,8,3784-3797,2018,IEEE.
• Andrea Dal Pozzolo. Adaptive Machine learning for credit card fraud detection ULB MLG PhD thesis (supervised by G. Bontempi).
• Fabrizio Carcillo, Andrea Dal Pozzolo, Yann-Aël Le Borgne, Olivier Caelen, Yannis Mazzer, Gianluca Bontempi. Scarff: a scalable framework for streaming credit card fraud detection with Spark, Information fusion,41, 182-194,2018,Elsevier.
• Fabrizio Carcillo, Yann-Aël Le Borgne, Olivier Caelen, Gianluca Bontempi. Streaming active learning strategies for real-life credit card fraud detection: assessment and visualization, International Journal of Data Science and Analytics, 5,4,285-300,2018,Springer International Publishing.
• Bertrand Lebichot, Yann-Aël Le Borgne, Liyun He, Frederic Oblé, Gianluca Bontempi Deep-Learning Domain Adaptation Techniques for Credit Cards Fraud Detection, INNSBDDL 2019: Recent Advances in Big Data and Deep Learning, pp 78-88, 2019.
• Fabrizio Carcillo, Yann-Aël Le Borgne, Olivier Caelen, Frederic Oblé, Gianluca Bontempi *Combining Unsupervised and Supervised...

Credit scoring models play a crucial role for financial institutions in evaluating borrower risk and sustaining profitability. Logistic regression is widely used in credit scoring due to its robustness, interpretability, and computational efficiency; however, its predictive power decreases when applied to complex or non-linear datasets, resulting in reduced accuracy. In contrast, tree-based machine learning models often provide enhanced predictive performance but struggle with interpretability. Furthermore, imbalanced class distributions, which are prevalent in credit scoring, can adversely impact model accuracy and robustness, as the majority class tends to dominate. Despite these challenges, research that comprehensively addresses both the predictive performance and explainability aspects within the credit scoring domain remains limited. This paper introduces the Non-pArameTric oversampling approach for Explainable credit scoring (NATE), a framework designed to address these challenges by combining oversampling techniques with tree-based classifiers to enhance model performance and interpretability. NATE incorporates class balancing methods to mitigate the impact of imbalanced data distributions and integrates interpretability features to elucidate the model’s decision-making process. Experimental results show that NATE substantially outperforms traditional logistic regression in credit risk classification, with improvements of 19.33% in AUC, 71.56% in MCC, and 85.33% in F1 Score. Oversampling approaches, particularly when used with gradient boosting, demonstrated superior effectiveness compared to undersampling, achieving optimal metrics of AUC: 0.9649, MCC: 0.8104, and F1 Score: 0.9072. Moreover, NATE enhances interpretability by providing detailed insights into feature contributions, aiding in understanding individual predictions. These findings highlight NATE’s capability in managing class imbalance, improving predictive performance, and enhancing model interpretability, demonstrating its potential as a reliable and transparent tool for credit scoring applications.

Context

Uplift modeling is an important yet novel area of research in machine learning which aims to explain and to estimate the causal impact of a treatment at the individual level. In the digital advertising industry, the treatment is exposure to different ads and uplift modeling is used to direct marketing efforts towards users for whom it is the most efficient . The data is a collection collection of 13 million samples from a randomized control trial, scaling up previously available datasets by a healthy 590x factor.

###
###

Content

The dataset was created by The Criteo AI Lab .The dataset consists of 13M rows, each one representing a user with 12 features, a treatment indicator and 2 binary labels (visits and conversions). Positive labels mean the user visited/converted on the advertiser website during the test period (2 weeks). The global treatment ratio is 84.6%. It is usual that advertisers keep only a small control population as it costs them in potential revenue.

Following is a detailed description of the features:

• f0, f1, f2, f3, f4, f5, f6, f7, f8, f9, f10, f11: feature values (dense, float)
• treatment: treatment group (1 = treated, 0 = control)
• conversion: whether a conversion occured for this user (binary, label)
• visit: whether a visit occured for this user (binary, label)
• exposure: treatment effect, whether the user has been effectively exposed (binary)

###

Context

Uplift modeling is an important yet novel area of research in machine learning which aims to explain and to estimate the causal impact of a treatment at the individual level. In the digital advertising industry, the treatment is exposure to different ads and uplift modeling is used to direct marketing efforts towards users for whom it is the most efficient . The data is a collection collection of 13 million samples from a randomized control trial, scaling up previously available datasets by a healthy 590x factor.

###
###

Content

The dataset was created by The Criteo AI Lab .The dataset consists of 13M rows, each one representing a user with 12 features, a treatment indicator and 2 binary labels (visits and conversions). Positive labels mean the user visited/converted on the advertiser website during the test period (2 weeks). The global treatment ratio is 84.6%. It is usual that advertisers keep only a small control population as it costs them in potential revenue.

Following is a detailed description of the features:

• f0, f1, f2, f3, f4, f5, f6, f7, f8, f9, f10, f11: feature values (dense, float)
• treatment: treatment group (1 = treated, 0 = control)
• conversion: whether a conversion occured for this user (binary, label)
• visit: whether a visit occured for this user (binary, label)
• exposure: treatment effect, whether the user has been effectively exposed (binary)

###

Starter Kernels

• HistGradientBoostingClassifier Base Model

Acknowledgement

The data provided for paper: "A Large Scale Benchmark for Uplift Modeling"

https://s3.us-east-2.amazonaws.com/criteo-uplift-dataset/large-scale-benchmark.pdf

• Eustache Diemert CAIL e.diemert@criteo.com
• Artem Betlei CAIL & Université Grenoble Alpes a.betlei@criteo.com
• Christophe Renaudin CAIL c.renaudin@criteo.com
• Massih-Reza Amini Université Grenoble Alpes massih-reza.amini@imag.fr

For privacy reasons the data has been sub-sampled non-uniformly so that the original incrementality level cannot be deduced from the dataset while preserving a realistic, challenging benchmark. Feature names have been anonymized and their values randomly projected so as to keep predictive power while making it practically impossible to recover the original features or user context.

Inspiration

We can foresee related usages such as but not limited to:

• Uplift modeling
• Interactions between features and treatment
• Heterogeneity of treatment

More Readings

• Supercharging customer touchpoints with uplift modeling
• CasualML
• PyLift

MORE DATASETs ...

Two synthetic datasets for binary classification, generated with the Random Radial Basis Function generator from WEKA. They are the same shape and size (104.952 instances, 185 attributes), but the "balanced" dataset has 52,13% of its instances belonging to class c0, while the "unbalanced" one only has 4,04% of its instances belonging to class c0. Therefore, this set of datasets is primarily meant to study how class balance influences the behaviour of a machine learning model.

Summary table: Oversampling techniques using SMOTE, ADASYN, and weighted rare classes.