Pith. sign in

REVIEW 3 major objections 5 minor 73 references

A school's average socioeconomic level, not individual traits, is the strongest predictor of a student's SAEB exam performance, this machine-learning study concludes.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

On 6.48 million Brazilian students, a Random Forest model predicted above/below-average proficiency with 90.2% accuracy, and the school's average socioeconomic level was the most influential predictor.

T0 review reviewed 2026-08-04 challenge →

load-bearing objection Solid data-engineering exercise and a confirmed known result, but the headline metrics and SHAP ranking rest on an invalid random split with school-level features, so the numbers aren't trustworthy as submitted. the 3 major comments →

arxiv 2510.22266 v2 pith:DH4IF2EP submitted 2025-10-25 cs.LG cs.AIcs.CY

A Multi-level Analysis of Factors Associated with Student Performance: A Machine Learning Approach to the SAEB Microdata

Rodrigo Tertulino , Ricardo Almeida This is my paper
classification cs.LG cs.AIcs.CY
keywords student performance predictionSAEB microdataeducational data miningschool socioeconomic levelRandom ForestSHAPmulti-level analysisBrazilian basic education
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that academic achievement in Brazilian basic education is a systemic phenomenon: the context a student learns in matters more than any isolated personal characteristic. Using 6,482,168 records from the national SAEB assessment, the authors integrate student, teacher, school, and principal data into one model. A Random Forest classifier reaches 90.2% accuracy in predicting whether a student scores above or below average, and SHAP attribution shows the school's average socioeconomic level is the dominant feature. If this is right, educational policy should aim at reducing disparities between schools rather than targeting students one by one.

Core claim

On the paper's own terms, the discovery is that the school's ecosystem — captured chiefly by the school's average socioeconomic level, the share of teachers with adequate training, and student participation rate — predicts student proficiency more strongly than any student-level characteristic. The authors demonstrate this by training four tree-based ensembles on a feature set refined from 98 candidates to 44 via the Boruta algorithm and then to 17 with specialist input, finding Random Forest's 90.2% accuracy and 96.7% AUC dominate untuned gradient boosting, and then showing with SHAP that school-level variables sit at the top of the importance ranking.

What carries the argument

The carrying mechanism is the multi-level table: four SAEB microdata files (student socioeconomic questionnaire, teacher profile, school indicators, principal questionnaire) joined by school and class IDs to make one hierarchical dataset. On top of that sits a Random Forest classifier, and the evidence for the systemic claim comes from SHAP (Shapley Additive Explanations), a model-agnostic method that attributes each prediction to its input features, which lets the authors rank the school's average socioeconomic level first.

Load-bearing premise

The load-bearing premise is that randomly splitting students into train and test while keeping school-level aggregates as features does not inflate accuracy; if the model is only ever tested on students from schools it already saw in training, school-level memorization could masquerade as a systemic effect.

What would settle it

Rerun the pipeline with a grouped split that places all students from a school in either training or testing, then compare accuracy, AUC, and the SHAP ranking of the school socioeconomic level. If accuracy drops toward the boosting models' roughly 63% or the school-level features lose their top rank, the central systemic conclusion is an artifact of within-school leakage.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If the systemic claim holds, funding formulas and teacher-allocation policies should weight school socioeconomic context heavily.
  • The model's balanced precision and recall across above- and below-average students means it can serve as a diagnostic, flagging which students are likely to underperform and which school-level drivers push them there.
  • The dominance of school-level indicators over individual teacher experience and degree implies that improving school composition and participation may yield larger gains than isolated professional-development programs.
  • Applying the same SHAP pipeline to later SAEB editions would let policymakers trace whether investments in teacher training shift feature importance over time.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Inference: because the data split randomly assigns students rather than schools, and school-level aggregates are features, part of the reported accuracy may come from the model recognizing individual schools; a leave-one-school-out validation is the test that would tell whether the systemic finding generalizes to unseen schools.
  • Inference: the school socioeconomic level is itself an average of students' socioeconomic scores, so the paper's ranking partly measures the predictive power of aggregating individual background up to the school level, not necessarily a separate institutional effect.
  • Inference: a natural extension the paper does not run — a multilevel or hierarchical model keeping school as a random effect — would quantify how much variance lies between schools versus within them, cleanly separating composition from context.
  • Inference: if school-level predictions prove robust out-of-school, the same four-layer design could be ported to other large-scale assessments, such as state-level exams, to guide decentralized education policy.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The manuscript integrates SAEB microdata from four levels (student, teacher, school, principal) and trains Random Forest, XGBoost, LightGBM, and CatBoost to classify students as above or below the mean composite proficiency. It reports that Random Forest achieves 90.2% accuracy and AUC 0.967, and SHAP analysis identifies the school's average socioeconomic level as the single most important predictor, leading the authors to conclude that academic performance is a systemic phenomenon shaped by the school ecosystem. The analysis uses 6,482,168 student records after listwise deletion and a 17-feature subset selected by Boruta and specialist consultation.

Significance. If the validation were sound, the paper would offer a useful large-scale, multi-level description of associations between school context and student achievement in Brazilian basic education, with substantial policy relevance. The manuscript's strengths are its integration of four SAEB data sources, its national sample size, and its attempt to move beyond prediction with SHAP. The data are public and the pipeline is described in algorithmic form. However, the headline accuracy and the central feature-importance claim are not currently supported because the evaluation uses a random student-level split while retaining school-level features, and the top predictor is partly an aggregate of the same students' socioeconomic responses. These are validity issues that affect the central conclusion, not merely presentation.

major comments (3)
  1. [§4.4, Algorithm 1 line 13, §5.1, Table 5] The random 80/20 student-level split with school-level features in the feature set creates group leakage. NIVEL SOCIO ECONOMICO, PC FORMACAO DOCENTE FINAL, and TAXA PARTICIPACAO are constant for all students in a school; with most schools appearing in both training and test sets, the model can learn a mapping from a school's aggregate features to that school's mean proficiency and apply it to held-out students from the same school. This inflates the reported 90.2% accuracy and AUC, and it inflates SHAP importance for school-level features, because those features are the vehicle for within-school memorization. Please re-evaluate with school-level grouped cross-validation (e.g., GroupKFold by school ID) and/or with school-level features removed, and report whether the performance and the feature-importance ranking change.
  2. [§5.3, Table 5, footnote 1] The top predictor, 'school's average socioeconomic level,' is an INEP composite calculated from the average of the same students' individual socioeconomic questionnaire responses. Thus the aggregate feature is a group-mean re-coding of individual-level SES, so the conclusion that systemic factors dominate individual characteristics is partly built into the measurement. To support the claim, the analysis should enter individual-level SES variables and school-level aggregates separately and compare their SHAP values after proper group-level validation.
  3. [§4.5] The 'final subset of 17 features' selected in consultation with specialists is never enumerated. Table 5 and Figure 4 show only a subset of features, and the mapping from the 44 Boruta-confirmed features to the final 17 is not given. Because feature selection occurs before model training and interpretation, the undisclosed subset directly affects the reported importance hierarchy and makes the analysis non-reproducible. Please provide the full list of the 17 features, their source level, and the selection criteria/protocol.
minor comments (5)
  1. [§5.1 vs. Abstract and Table 3] The text reports 90.17% accuracy and 0.9669 AUC while the abstract and conclusions use 90.2% and 96.7%; these should be consistent. Also, §5.1 says Random Forest is 'substantially superior' while §6.1 calls the advantage 'marginal'; align the wording.
  2. [§5.1, Figure 2] The terms 'true negative rate 94.3%' and 'true positive rate 92.8%' are nonstandard for this confusion-matrix description. Clarify whether these are class-specific recall values for the 'Below Average' and 'Above Average' classes, or compute standard sensitivity/specificity.
  3. [§5.3, Figure 4] The SHAP summary is based on a 20,000-record sample, but the sampling method is not described. State how the sample was drawn, whether school structure was preserved, and whether the sample size was chosen for computational reasons.
  4. [§7.1] The limitations section mentions listwise deletion and cross-sectional design but does not acknowledge the validation flaw of student-level splitting with school-level features. Add this as a limitation and, more importantly, correct it in the analysis.
  5. [Code Availability] The paper says code will be released 'upon publication'; for a reproducibility review, it would be helpful to provide the code and the exact preprocessing/feature-selection scripts in the current version, especially because the 17-feature subset is not enumerated.

Circularity Check

1 steps flagged

Student-level random split with school-level features makes the reported accuracy and SHAP 'systemic dominance' partly a within-school reconstruction.

specific steps
  1. fitted input called prediction [Algorithm 1, line 13; Section 4.2; Section 5.2/5.3]
    ""X train, Xtest, ytrain, ytest←SplitData(X, y, testsize= 0.2)" (Algorithm 1); "To prevent data leakage, all identifier columns were explicitly removed from the feature set prior to training." (Section 4.2); "NIVEL SOCIO ECONOMICO ... is derived from the average of individual student socioeconomic scores" (Section 5.2 footnote)."

    The split is random at the student level, not grouped by school, while the feature set includes school-level aggregates (NIVEL SOCIO ECONOMICO, PC FORMACAO DOCENTE FINAL, TAXA PARTICIPACAO) that are constant within a school. Under Algorithm 1's SplitData(X, y, testsize=0.2), most schools appear in both train and test. A tree model can learn a mapping from these school-level features to the school's mean proficiency using training students, then apply that mapping to test students from the same school. The reported 90.2% accuracy/AUC and the top SHAP rank of school-level features are thus partly a reconstruction of school-level outcomes, not an independent estimate of systemic generalization. The 'prediction' for a held-out student is statistically forced by the school-level fit from that s

full rationale

The paper contains no equation-level derivation; its central claims are empirical. I examined the feature-construction and validation chain for reductions. The only load-bearing circularity is the evaluation protocol: Algorithm 1 splits students randomly, while the feature set includes school-level aggregates constant within schools. Consequently, test students from a school that also contributed training rows can be predicted from the school-level mapping learned on those same rows; the reported accuracy/AUC and SHAP dominance of school-level features are partly within-school reconstructions rather than evidence for the systemic-dominance conclusion. The Limitations section (7.1) flags listwise deletion and cross-sectional design, but not this group-leakage issue. Separately, NIVEL SOCIO ECONOMICO is an aggregate of individual student questionnaire responses, so the 'systemic vs individual' contrast is partly a comparison of an aggregated version of individual SES with unaggregated individual variables; this compounds interpretation but is not itself a tautology because the aggregate is not built from the proficiency target. No self-citation chain or imported uniqueness theorem is load-bearing. Score 6: the central quantitative claim partially reduces to the leakage, but the paper's descriptive finding that school SES correlates with achievement has independent support in the literature.

Axiom & Free-Parameter Ledger

3 free parameters · 6 axioms · 0 invented entities

The paper introduces no new theoretical entities or physical parameters. The contributed quantities are data-derived: the binary target threshold, the expert-selected 17-feature subset, and the SHAP sample. The analysis leans on standard ML axioms and on domain assumptions about SAEB validity, the non-leakage of random splitting, and the interpretability of feature importance.

free parameters (3)
  • Binary target threshold = mean of composite Portuguese+Mathematics proficiency in the analyzed sample
    The above/below-average class boundary is computed from the same data the model is trained on; results depend on this data-derived cutoff (§4.4).
  • Final feature subset (17 of 44 confirmed features) = 17 features selected 'in consultation with specialists'
    The final set is not enumerated, and the selection uses domain expertise plus Boruta results; this is a hand-chosen component of the pipeline (§4.5).
  • SHAP analysis sample size = 20,000 records
    The global SHAP summary is computed on a 20,000-record sample; representativeness is not discussed (§5.3, Figure 4 caption).
axioms (6)
  • domain assumption SAEB proficiency scores and the contextual questionnaire responses validly measure student performance and its correlates.
    The whole analysis treats INEP's proficiency scales and self-reported student/teacher/principal questionnaire items as reliable measures (§3, §4.1).
  • ad hoc to paper Binarizing the target at the mean composite proficiency of the analyzed sample yields meaningful above/below-average classes.
    The class boundary is data-derived from the same cohort; §4.4. Results depend on this threshold.
  • ad hoc to paper Random student-level 80/20 splitting does not leak school-level information through school-level features.
    The model includes school socioeconomic level and other school aggregates; the same school can appear in both train and test. This is assumed without validation (§4.2, §4.4).
  • domain assumption Feature importance/SHAP values from the trained classifier can be interpreted as the factors 'associated with' student performance.
    The paper shifts from prediction to explanation (§5.3) but does not validate SHAP faithfulness or causal identifiability; §4.6 acknowledges association, not causality.
  • domain assumption Listwise deletion of missing records leaves a representative sample (or introduces only minor selection bias).
    The paper acknowledges possible selection bias (§7.1); the analyzed 6.48M records are a subset of the full SAEB data.
  • domain assumption The INEP school socioeconomic level indicator (NIVEL SOCIO ECONOMICO) is a valid measure of school-level socioeconomic composition.
    This variable is the central predictor; its construction is referenced to INEP 2023 but not independently audited (§3, Table 5, Appendix A).

reviewed 2026-08-04 · how reviews work

0 comments
Cite this review

Pith. "Pith review of A Multi-level Analysis of Factors Associated with Student Performance: A Machine Learning Approach to the SAEB Microdata." pith.science (2026). https://pith.science/paper/DH4IF2EP

@misc{pith2026251022266,
  author       = {Pith},
  title        = {Pith review of: A Multi-level Analysis of Factors Associated with Student Performance: A Machine Learning Approach to the SAEB Microdata},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DH4IF2EP}},
  note         = {Machine review of arXiv:2510.22266}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Identifying the factors that influence student performance in basic education is a central challenge for formulating effective public policies in Brazil. This study introduces a multi-level machine learning approach to classify the proficiency of 9th-grade and high school students using microdata from the System of Assessment of Basic Education (SAEB). Our model uniquely integrates four data sources: student socioeconomic characteristics, teacher professional profiles, school indicators, and principal management profiles. A comparative analysis of four ensemble algorithms confirmed the superiority of a Random Forest model, which achieved 90.2% accuracy and an Area Under the Curve (AUC) of 96.7%. To move beyond prediction, we applied Explainable AI (XAI) using SHAP, which revealed that the school's average socioeconomic level is the most dominant predictor, demonstrating that systemic factors have a greater impact than individual characteristics in isolation. The primary conclusion is that academic performance is a systemic phenomenon deeply tied to the school's ecosystem. This study provides a data-driven, interpretable tool to inform policies aimed at promoting educational equity by addressing disparities between schools.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

73 extracted references · 29 canonical work pages

  1. [1]

    , Al-Samarraie , H

    barticle Aldowah , H. , Al-Samarraie , H. , Fauzy , W.M.H. : Educational data mining and learning analytics for insightful information in higher education: A review . Education and Information Technologies 24 ( 2 ), 1179 -- 1212 ( 2019 ) 10.1016/J.TELE.2019.01.007 barticle

    doi:10.1016/j.tele.2019.01.007 2019

  2. [2]

    , Abduljabbar , Z

    barticle Ali , Z. , Abduljabbar , Z. , Tahir , H. , Sallow , A. , Almufti , S. : extreme gradient boosting algorithm with machine learning: A review . Academic Journal of Nawroz University 12 , 320 -- 334 ( 2023 ) 10.25007/ajnu.v12n2a1612 barticle

    work page doi:10.25007/ajnu.v12n2a1612 2023

  3. [3]

    , Berrada , M

    barticle Adadi , A. , Berrada , M. : Peeking inside the black-box: A survey on explainable artificial intelligence (xai) . IEEE Access 6 , 52138 -- 52160 ( 2018 ) 10.1109/ACCESS.2018.2870052 barticle

    arXiv 2018

  4. [4]

    , G \'a mez , M

    barticle Alfaro , E. , G \'a mez , M. , Garc \' a , N. : Combining bagging and boosting . Journal of Statistical Computation and Simulation 83 ( 4 ), 721 -- 740 ( 2013 ) 10.1080/00949655.2011.627702 barticle

    arXiv 2013

  5. [5]

    , Karabatak , S

    bchapter Alan , A. , Karabatak , S. , Karabatak , M. : The role of machine learning in distance education . In: 2025 13th International Symposium on Digital Forensics and Security (ISDFS) , pp. 1 -- 6 ( 2025 ). 10.1109/ISDFS65363.2025.11012110 bchapter

    arXiv 2025

  6. [6]

    , Soares , J.F

    barticle Alves , M.T.G. , Soares , J.F. : As desigualdades educacionais no brasil: o que mostra o saeb/prova brasil? Cadernos de Pesquisa 43 ( 148 ), 56 -- 79 ( 2013 ) 10.1590/S0100-15742013000100004 barticle

    work page doi:10.1590/s0100-15742013000100004 2013

  7. [7]

    , Sehgal , R

    barticle Anand , N. , Sehgal , R. , Anand , S. , Kaushik , A. : Feature selection on educational data using boruta algorithm . Int. J. Comput. Intell. Stud. 10 ( 1 ), 27 -- 35 ( 2021 ) 10.1504/ijcistudies.2021.113826 barticle

    arXiv 2021

  8. [8]

    , Zaki , N

    barticle Albreiki , B. , Zaki , N. , Alashwal , H. : A systematic literature review of student' performance prediction using machine learning techniques . Education Sciences 11 ( 9 ), 552 ( 2021 ) 10.3390/educsci11090552 barticle

    doi:10.3390/educsci11090552 2021

  9. [9]

    , Artar , M

    barticle Balcıoğlu , Y.S. , Artar , M. : Predicting academic performance of students with machine learning . Information Development 41 ( 3 ), 896 -- 915 ( 2023 ) 10.1177/02666669231213023 barticle

    work page doi:10.1177/02666669231213023 2023

  10. [10]

    : The world bank and unesco institute for statistics (uis)

    botherref Bank , W. : The world bank and unesco institute for statistics (uis). the state of global learning poverty: 2022 update. Technical report, The World Bank, Washington, D.C. (2022). https://www.worldbank.org/en/topic/education/publication/state-of-global-learning-poverty botherref

    2022

  11. [11]

    , B \`a ssolo , \'A

    barticle Bonamino , A. , B \`a ssolo , \'A . , Franco , C. : Os caminhos da avalia c \ a o em larga escala no brasil: o caso do saeb . Meta: Avalia c \ a o 2 ( 4 ), 113 -- 134 ( 2010 ) 10.22347/2175-2753v2i4.46 barticle

    work page doi:10.22347/2175-2753v2i4.46 2010

  12. [12]

    , Csorgo , A

    barticle Bentejac , C. , Csorgo , A. , Martinez , G.J. : A comparative analysis of gradient boosting algorithms . Artificial Intelligence Review 54 , 1937 -- 1967 ( 2021 ) 10.1007/s10462-020-09896-5 barticle

    doi:10.1007/s10462-020-09896-5 1937

  13. [13]

    , Gintis , H

    bbook Bowles , S. , Gintis , H. : Schooling in Capitalist America: Educational Reform and the Contradictions of Economic Life . Basic Books , New York ( 1976 ) bbook

    1976

  14. [14]

    , Guzm \'a n Rinc \'o n , A

    botherref Barrag \'a n Moreno , S.P. , Guzm \'a n Rinc \'o n , A. : Digital divide as an explanatory variable for dropout in higher education. Int. J. Educ. Technol. High. Educ. 22(1) (2025) 10.1186/s41239-025-00550-0 botherref

    work page doi:10.1186/s41239-025-00550-0 2025

  15. [15]

    , Gonz \'a lez T \'a mara , L

    bchapter Barrag \'a n Moreno , S.P. , Gonz \'a lez T \'a mara , L. : Complexities of student dropout in higher education: a multidimensional analysis . In: Frontiers in Education , vol. 9 , p. 1461650 ( 2024 ). Frontiers Media SA bchapter

    2024

  16. [16]

    : The Forms of Capital

    bbook Bourdieu , P. : The Forms of Capital . Greenwood Press , New York ( 1986 ). In J. G. Richardson (Ed.), Handbook of Theory and Research for the Sociology of Education (pp. 241-258). bbook

    1986

  17. [17]

    : Random forests

    barticle Breiman , L. : Random forests . Machine learning 45 ( 1 ), 5 -- 32 ( 2001 ) 10.1023/A:1010933404324 barticle

    doi:10.1023/a:1010933404324 2001

  18. [18]

    , Dutra , B.L

    barticle Carneiro , M.G. , Dutra , B.L. , Paiva , J.G.S. , Gabriel , P.H.R. , Araújo , R.D. : Educational data mining to support identification and prevention of academic retention and dropout: a case study in introductory programming . Revista Brasileira de Informática na Educação - RBIE 30 , 379 -- 395 ( 2022 ) 10.5753/rbie.2022.2518 barticle

    arXiv 2022

  19. [19]

    , Ezeife , C.I

    bchapter Chaturvedi , R. , Ezeife , C.I. : Predicting student performance in an its using task-driven features . In: 2017 IEEE International Conference on Computer and Information Technology (CIT) , pp. 334 -- 339 ( 2017 ). 10.1109/CIT.2017.65 . IEEE bchapter

    work page doi:10.1109/cit.2017.65 2017

  20. [20]

    , Guestrin , C

    bchapter Chen , T. , Guestrin , C. : Xgboost: A scalable tree boosting system . In: Proceedings of the 22nd Acm Sigkdd International Conference on Knowledge Discovery and Data Mining , pp. 785 -- 794 ( 2016 ). 10.1145/2939672.2939785 bchapter

    arXiv 2016

  21. [21]

    , Khoroshavin , A

    barticle Chassignol , M. , Khoroshavin , A. , Klimova , A. , Bilyatdinova , A. : Explainable artificial intelligence for education and training . Procedia Computer Science 136 , 476 -- 480 ( 2018 ) 10.1016/j.procs.2018.08.280 barticle

    work page doi:10.1016/j.procs.2018.08.280 2018

  22. [22]

    , Pereira , E.M

    botherref Carvalho , D.V. , Pereira , E.M. , Cardoso , J.S. : Machine learning interpretability: A survey on methods and metrics. Electronics 8(8) (2019) 10.3390/electronics8080832 botherref

    doi:10.3390/electronics8080832 2019

  23. [23]

    , Thompsen Primo , T

    barticle Colpo , M.P. , Thompsen Primo , T. , Aguiar , M.S. , Cechinel , C. : Educational data mining for dropout prediction: Trends, opportunities, and challenges . Revista Brasileira de Informática na Educação 32 , 220 -- 256 ( 2024 ) 10.5753/rbie.2024.3559 barticle

    arXiv 2024

  24. [24]

    , Dia , N.J

    barticle Deleña , R.D. , Dia , N.J. , Sacayan , R.R. , Sieras , J.C. , Khalid , S.A. , Macatotong , A.H.T. , Gulam , S.B. : Predicting student retention: A comparative study of machine learning approach utilizing sociodemographic and academic factors . Systems and Soft Computing 7 , 200352 ( 2025 ) 10.1016/j.sasc.2025.200352 barticle

    arXiv 2025

  25. [25]

    , Ershov , V

    bchapter Dorogush , A.V. , Ershov , V. , Gulin , A. : Catboost: gradient boosting with categorical features support . In: Proceedings of the 2nd International Workshop on Data Science ( 2018 ) bchapter

    2018

  26. [26]

    : Ensemble methods in machine learning

    bchapter Dietterich , T.G. : Ensemble methods in machine learning . In: Multiple Classifier Systems , pp. 1 -- 15 . Springer , Berlin, Heidelberg ( 2000 ) bchapter

    2000

  27. [27]

    , Rad , P

    botherref Das , A. , Rad , P. : Opportunities and challenges in explainable artificial intelligence (XAI): A survey. CoRR abs/2006.11371 (2020) https://arxiv.org/abs/2006.11371 2006.11371 botherref

    Pith/arXiv arXiv 2006

  28. [28]

    : O saeb - sistema de avaliação da educação básica: potencialidades, problemas e desafios

    botherref Franco , C. : O saeb - sistema de avaliação da educação básica: potencialidades, problemas e desafios. Revista Brasileira de Educação (17), 127--133 (2001) 10.1590/S1413-24782001000200010 botherref

    work page doi:10.1590/s1413-24782001000200010 2001

  29. [29]

    : Visible Learning: A Synthesis of Over 800 Meta-Analyses Relating to Achievement , ( 2009 )

    bbook Hattie , J. : Visible Learning: A Synthesis of Over 800 Meta-Analyses Relating to Achievement , ( 2009 ). 10.4324/9780203887332 bbook

    work page doi:10.4324/9780203887332 2009

  30. [30]

    , Appiah , O

    barticle Issah , I. , Appiah , O. , Appiahene , P. , Inusah , F. : A systematic review of the literature on machine learning application of determining the attributes influencing academic performance . Decision Analytics Journal 7 , 100204 ( 2023 ) 10.1016/j.dajour.2023.100204 barticle

    arXiv 2023

  31. [31]

    , Beikmohammadi , A

    botherref Imani , M. , Beikmohammadi , A. , Arabnia , H.R. : Comprehensive analysis of random forest and xgboost performance with smote, adasyn, and gnus under varying imbalance levels. Technologies 13(3) (2025) 10.3390/technologies13030088 botherref

    work page doi:10.3390/technologies13030088 2025

  32. [32]

    : Exploring the role of explainable artificial intelligence(xai) in adaptive learning systems

    bchapter Idrizi , E. : Exploring the role of explainable artificial intelligence(xai) in adaptive learning systems . In: Proceedings of the Cognitive Models and Artificial Intelligence Conference . AICCONF '24 , pp. 100 -- 105 . Association for Computing Machinery , New York, NY, USA ( 2024 ). 10.1145/3660853.3660877 bchapter

    arXiv 2024

  33. [33]

    Technical report, Ministério da Educação, Brasília, DF (2023)

    botherref INEP : Instituto nacional de estudos e pesquisas educacionais anísio teixeira (inep) - indicador de nível socioeconômico do saeb 2021: Nota técnica. Technical report, Ministério da Educação, Brasília, DF (2023). https://www.gov.br/inep/pt-br/centrais-de-conteudo/acervo-linha-editorial/publicacoes-institucionais/avaliacoes-e-exames-da-educacao-ba...

    2021

  34. [34]

    https://www.gov.br/inep/pt-br/acesso-a-informacao/dados-abertos/microdados/saeb

    botherref INEP : Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira (INEP) - Microdados do Sistema de Avaliação da Educação Básica (Saeb). https://www.gov.br/inep/pt-br/acesso-a-informacao/dados-abertos/microdados/saeb. Accessed: 2/8/25 (2025) botherref

    2025

  35. [35]

    , Choi , S

    barticle Jang , Y. , Choi , S. , Jung , H. , Kim , H. : Practical early prediction of students’ performance using machine learning and explainable ai . Education and Information Technologies 27 ( 9 ), 12855 -- 12889 ( 2022 ) 10.1007/s10639-022-11120-6 barticle

    work page doi:10.1007/s10639-022-11120-6 2022

  36. [36]

    , Engelke , A

    botherref Jungmair , M. , Engelke , A. , Giceva , J. : Hipy: Extracting high-level semantics from python code for data processing. Proc. ACM Program. Lang. 8(OOPSLA2) (2024) 10.1145/3689737 botherref

    doi:10.1145/3689737 2024

  37. [37]

    , Kiraz , S

    botherref Kesgin , K. , Kiraz , S. , Kosunalp , S. , Stoycheva , B. : Beyond performance: Explaining and ensuring fairness in student academic performance prediction with machine learning. Applied Sciences 15(15) (2025) 10.3390/app15158409 botherref

    work page doi:10.3390/app15158409 2025

  38. [38]

    , Meng , Q

    bchapter Ke , G. , Meng , Q. , Finley , T. , Wang , T. , Chen , W. , Ma , W. , Ye , Q. , Liu , T.-Y. : Lightgbm: A highly efficient gradient boosting decision tree . In: Advances in Neural Information Processing Systems 30 ( 2017 ) bchapter

    2017

  39. [39]

    , Saltarelli , A.J

    barticle Kizilcec , R.F. , Saltarelli , A.J. : Equity and learning analytics: a research agenda . Journal of Learning Analytics 7 ( 3 ), 11 -- 20 ( 2020 ) 10.18608/jla.2020.73.2 barticle

    work page doi:10.18608/jla.2020.73.2 2020

  40. [40]

    , Lee , S.-I

    bchapter Lundberg , S. , Lee , S.-I. : A unified approach to interpreting model predictions . ( 2017 ). 10.48550/arXiv.1705.07874 bchapter

    Pith/arXiv arXiv doi:10.48550/arxiv.1705.07874 2017

  41. [41]

    , Lee , S.-I

    bchapter Lundberg , S.M. , Lee , S.-I. : A unified approach to interpreting model predictions . In: Proceedings of the 31st International Conference on Neural Information Processing Systems . NIPS'17 , pp. 4768 -- 4777 . Curran Associates Inc. , Red Hook, NY, USA ( 2017 ) bchapter

    2017

  42. [42]

    , Nguyen , A.J

    barticle Larson , K.E. , Nguyen , A.J. , Orozco Solis , M.G. , Humphreys , A. , Bradshaw , C.P. , Lindstrom Johnson , S. : A systematic literature review of school climate in low and middle income countries . International Journal of Educational Research 102 , 101606 ( 2020 ) 10.1016/j.ijer.2020.101606 barticle

    arXiv 2020

  43. [43]

    , Porter , L

    barticle Liebowitz , D.D. , Porter , L. : The effect of principal behaviors on student, teacher, and school outcomes: A systematic review and meta-analysis of the empirical literature . Review of Educational Research 89 ( 5 ), 785 -- 827 ( 2019 ) 10.3102/0034654319866133 https://arxiv.org/abs/https://doi.org/10.3102/0034654319866133 https://doi.org/10.310...

    work page doi:10.3102/0034654319866133 2019

  44. [44]

    , Weininger , E.B

    barticle Lareau , A. , Weininger , E.B. : Revisiting cultural capital: The role of the family in the digital age . Journal of Cultural Sociology 17 ( 1 ), 3 -- 25 ( 2023 ) 10.1177/17499755221142517 barticle

    work page doi:10.1177/17499755221142517 2023

  45. [45]

    , Deroy , A

    botherref Maity , S. , Deroy , A. : Human-Centric eXplainable AI in Education (2024). https://arxiv.org/abs/2410.19822 botherref

    Pith/arXiv arXiv 2024

  46. [46]

    , Dehghani , T

    barticle Mastour , H. , Dehghani , T. , Moradi , E. , Eslami , S. : Explainable artificial intelligence for predicting medical students' performance in comprehensive assessments . Scientific Reports 15 ( 1 ), 23752 ( 2025 ) 10.1038/s41598-025-07460-1 barticle

    work page doi:10.1038/s41598-025-07460-1 2025

  47. [47]

    : Explanation in artificial intelligence: Insights from the social sciences

    barticle Miller , T. : Explanation in artificial intelligence: Insights from the social sciences . Artificial Intelligence 267 , 1 -- 38 ( 2019 ) 10.1016/j.artint.2018.07.007 barticle

    doi:10.1016/j.artint.2018.07.007 2019

  48. [48]

    , Oliveira , D.A

    barticle Mazoni , A.R.G. , Oliveira , D.A. : Da distinção social à performatividade: escolas de excelência no brasil e no chile . Revista Educação e Políticas em Debate 12 ( 2 ), 600 -- 622 ( 2023 ) 10.14393/REPOD-v12n2a2023-69401 barticle

    work page doi:10.14393/repod-v12n2a2023-69401 2023

  49. [49]

    , Souza , S

    barticle Melo , C. , Souza , S. : Improving the prediction of school dropout with the support of the semi-supervised learning approach . iSys: Revista Brasileira de Sistemas de Informação 16 ( 1 ), 10 -- 11026 ( 2023 ) 10.5753/isys.2023.2852 barticle

    arXiv 2023

  50. [50]

    , Sood , K

    botherref Martinez , A.L.J. , Sood , K. , Mahto , R. : Early detection of at-risk students using machine learning. arXiv preprint arXiv:2412.09483 (2024) botherref

    Pith/arXiv arXiv 2024

  51. [51]

    PISA 2022 Results (Volume I): The State of Learning and Equity in Education

    bbook OECD : Organisation for Economic Co-operation and Development (OECD). PISA 2022 Results (Volume I): The State of Learning and Equity in Education . OECD Publishing , Paris ( 2023 ). 10.1787/53f23881-en bbook

    doi:10.1787/53f23881-en 2022

  52. [52]

    , Gulbandilar , E

    barticle Ozcan , G. , Gulbandilar , E. , Kocak , Y. : Prediction of compressive strengths of portland cement with random forest, support vector machine and gradient boosting models . Neural Comput. Appl. 37 ( 28 ), 23495 -- 23511 ( 2025 ) 10.1007/s00521-025-11536-4 barticle

    work page doi:10.1007/s00521-025-11536-4 2025

  53. [53]

    , Boulesteix , A.-L

    barticle Probst , P. , Boulesteix , A.-L. , Bischl , B. : Tunability: importance of hyperparameters of machine learning algorithms . J. Mach. Learn. Res. 20 ( 1 ), 1934 -- 1965 ( 2019 ) barticle

    1934

  54. [54]

    , Bortolotti , S

    bchapter Prado , C. , Bortolotti , S. , Barbetta , P. : Fatores associados ao desempenho em matemática dos estudantes do 9º ano do ensino fundamental, baseado no saeb 2017 . ( 2021 ). 10.29327/156288.1-9 bchapter

    work page doi:10.29327/156288.1-9 2017

  55. [55]

    , Fonseca , S.C

    barticle Pereira , F.D. , Fonseca , S.C. , Oliveira , E.H.T. , Oliveira , D.B.F. , Cristea , A.I. , Carvalho , L.S.G. : Deep learning for early performance prediction of introductory programming students: a comparative and explanatory study . Revista Brasileira de Informática na Educação - RBIE 28 , 723 -- 749 ( 2020 ) 10.5753/RBIE.2020.28.0.723 barticle

    work page doi:10.5753/rbie.2020.28.0.723 2020

  56. [56]

    , Wallace , M

    barticle Papadogiannis , I. , Wallace , M. , Karountzou , G. : Educational data mining: A foundational overview . Encyclopedia 4 ( 1 ), 1 -- 15 ( 2024 ) 10.3390/encyclopedia4010001 barticle

    work page doi:10.3390/encyclopedia4010001 2024

  57. [57]

    , Lourens , M

    bchapter Ravuri , A. , Lourens , M. , Aswini , S. , Nijhawan , G. , Zabibah , R.S. , Chandrashekar , R. : Improving personalized education: A machine learning method for flexible learning environments . In: 2023 10th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON) , vol. 10 , pp. 1715 -- 1720...

    arXiv 2023

  58. [58]

    , Lakshmi , K.P

    bchapter Raju , V.N.G. , Lakshmi , K.P. , Jain , V.M. , Kalidindi , A. , Padma , V. : Study the influence of normalization/transformation process on the accuracy of supervised classification . In: 2020 Third International Conference on Smart Systems and Inventive Technology (ICSSIT) , pp. 729 -- 735 ( 2020 ). 10.1109/ICSSIT48917.2020.9214160 bchapter

    arXiv 2020

  59. [59]

    , Soares , J.F

    botherref Rocha , E. , Soares , J.F. : O efeito da composi c \ a o socioecon \^o mica das escolas no desempenho dos alunos: uma an \'a lise dos dados do saeb 2019. Estudos em Avalia c \ a o Educacional 32 (2021) 10.18222/eae.v32.7562 botherref

    work page doi:10.18222/eae.v32.7562 2019

  60. [60]

    , Ventura , S

    barticle Romero , C. , Ventura , S. : Educational data mining: a review of the state of the art . IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews) 40 ( 6 ), 601 -- 618 ( 2010 ) 10.1109/TSMCC.2010.2053532 barticle

    arXiv 2010

  61. [61]

    , Balby , L

    bchapter Sales , A. , Balby , L. , Cajueiro , A. : Exploiting academic records for predicting student drop out: a case study in brazilian higher education . In: Proceedings of the 3rd Symposium on Knowledge Discovery, Mining and Learning , pp. 174 -- 181 ( 2015 ). 10.5753/kdmile.2015.6885 bchapter

    arXiv 2015

  62. [62]

    , Lin , Y.-R

    bchapter Sahebi , S. , Lin , Y.-R. , Brusilovsky , P. : Tensor factorization for student modeling and performance prediction in unstructured domain . ( 2016 ) bchapter

    2016

  63. [63]

    : Predictive analysis of students’ learning performance using data mining techniques: A comparative study of feature selection methods

    botherref Syed Mustapha , S.M.F.D. : Predictive analysis of students’ learning performance using data mining techniques: A comparative study of feature selection methods. Applied System Innovation 6(5) (2023) 10.3390/asi6050086 botherref

    work page doi:10.3390/asi6050086 2023

  64. [64]

    , Newman , A

    bchapter Shields , L. , Newman , A. , Satz , D. : Equality of Educational Opportunity . In: Zalta , E.N. , Nodelman , U. (eds.) The Stanford Encyclopedia of Philosophy , W inter 2023 edn. Metaphysics Research Lab, Stanford University , ??? ( 2023 ) bchapter

    2023

  65. [65]

    : O modelo de tr \^e s n \' veis: uma alternativa para a an \'a lise de dados de avalia c \ a o de sistemas educacionais

    barticle Soares , J.F. : O modelo de tr \^e s n \' veis: uma alternativa para a an \'a lise de dados de avalia c \ a o de sistemas educacionais . Ensaio: Avalia c \ a o e Pol \' ticas P \'u blicas em Educa c \ a o 13 ( 47 ), 149 -- 162 ( 2005 ) 10.1590/S0104-40362005000200003 barticle

    work page doi:10.1590/s0104-40362005000200003 2005

  66. [66]

    , Chen , Y

    bchapter Tang , X. , Chen , Y. : Adaptive education platform based on machine learning: A new way to improve the quality of higher education . In: 2024 International Conference on Interactive Intelligent Systems and Techniques (IIST) , pp. 310 -- 316 ( 2024 ). 10.1109/IIST62526.2024.00054 bchapter

    arXiv 2024

  67. [67]

    , Kappel , M.A.A

    barticle Teodoro , L.d.A. , Kappel , M.A.A. : Aplicação de técnicas de aprendizado de máquina para predição de risco de evasão escolar em instituições públicas de ensino superior no brasil . Revista Brasileira de Informática na Educação 28 , 838 -- 863 ( 2020 ) 10.5753/RBIE.2020.28.0.838 barticle

    work page doi:10.5753/rbie.2020.28.0.838 2020

  68. [68]

    , Chinmulgund , A

    bchapter Taufikin , Supa’At , Sharma , M. , Chinmulgund , A. , Kuanr , J. , Fatma , G. : The future of teaching: Exploring the integration of machine learning in higher education . In: 2024 International Conference on Knowledge Engineering and Communication Systems (ICKECS) , vol. 1 , pp. 1 -- 6 ( 2024 ). 10.1109/ICKECS61492.2024.10616421 bchapter

    arXiv 2024

  69. [69]

    , Development , O

    bbook Ulferts , H. , Development , O. , Hu , I. , Echazarra , A. : PISA 2022 Results (Volume II): Learning During – and From – Disruption , ( 2023 ). 10.1787/a97db61c-en bbook

    work page doi:10.1787/a97db61c-en 2022

  70. [70]

    , Minkovska , D

    barticle Ujkani , B. , Minkovska , D. , Hinov , N. : Course success prediction and early identification of at-risk students using explainable artificial intelligence . Electronics 13 ( 21 ), 4157 ( 2024 ) 10.3390/electronics13214157 barticle

    work page doi:10.3390/electronics13214157 2024

  71. [71]

    Global Education Monitoring Report 2020: Inclusion and Education: All Means All

    bbook UNESCO : United Nations Educational, Scientific and Cultural Organization (UNESCO). Global Education Monitoring Report 2020: Inclusion and Education: All Means All . UNESCO , Paris ( 2020 ). https://unesdoc.unesco.org/ark:/48223/pf0000373724 bbook

    2020

  72. [72]

    : School composition and contextual effects on student outcomes

    barticle Willms , J. : School composition and contextual effects on student outcomes . Teachers College Record 112 , 1008 -- 1037 ( 2010 ) 10.1177/016146811011200408 barticle

    work page doi:10.1177/016146811011200408 2010

  73. [73]

    , C evik , M

    barticle Ya g ci , A. , C evik , M. : Prediction of academic achievements of vocational and technical high school (vts) students in science courses through artificial neural networks (comparison of turkey and malaysia) . Education and Information Technologies 24 , 2741 -- 2761 ( 2019 ) 10.1007/s10639-019-09885-4 barticle

    work page doi:10.1007/s10639-019-09885-4 2019

This paper was first reviewed by deepseek-v4-flash on August 4, 2026.