We present the aggregated conformal predictor (ACP), an extension to the traditional inductive conformal prediction (ICP) where several inductive conformal predictors are applied on the same training set and their individual predictions are aggregated to form a single prediction on an example. The results from applying ACP on two pharmaceutical data sets (CDK5 and GNRHR) indicate that the ACP has advantages over traditional ICP. ACP reduces the variance of the prediction region estimates and improves efficiency. Still, it is more conservative in terms of validity than ICP, indicating that there is room for further improvement of efficiency without compromising validity.


Support Vector Machine Random Forest QSAR Model True Label Prediction Region 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    Eklund, M., Norinder, U., Boyer, S., Carlsson, L.: Application of Conformal Prediction in QSAR. In: Iliadis, L., Maglogiannis, I., Papadopoulos, H., Karatzas, K., Sioutas, S. (eds.) AIAI 2012, Part II. IFIP AICT, vol. 382, pp. 166–175. Springer, Heidelberg (2012)CrossRefGoogle Scholar
  2. 2.
    Eklund, M., Norinder, U., Boyer, S., Carlsson, L.: The application of conformal prediction to the drug discovery process. Annals of Mathematics and Artificial Intelligence (2013)Google Scholar
  3. 3.
    Vovk, V., Gammerman, A., Shafer, G.: Algorithmic Learning in a Random World, 1st edn. Springer (2005)Google Scholar
  4. 4.
    Vovk, V.: Cross-conformal predictors. Annals of Mathematics and Artificial Intelligence, 1–20 (2013)Google Scholar
  5. 5.
    Vovk, V.: Cross-conformal predictor. Working Paper 6 (2013),
  6. 6.
    Bühlmann, P., Yu, B.: Analyzing bagging. The Annals of Statistics 30(4), 927–961 (2002)MathSciNetCrossRefzbMATHGoogle Scholar
  7. 7.
    Ruppert, D.: Statistics and Data Analysis for Financial Engineering, 1st edn. Springer Texts in Statistics. Springer, Berlin (2010)Google Scholar
  8. 8.
    Vapnik, V.N.: Statistical learning theory, 1 edn. Wiley (1998)Google Scholar
  9. 9.
    Chang, C.C., Lin, C.J.: LIBSVM: A library for support vector machines. ACM Transactions on Intelligent Systems and Technology 2, 27:1–27:27 (2011), Software available at
  10. 10.
    Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., Duchesnay, E.: Scikit-learn: Machine learning in Python. Journal of Machine Learning Research 12, 2825–2830 (2011)zbMATHGoogle Scholar
  11. 11.
    Papadopoulos, H., Haralambous, H.: Reliable prediction intervals with regression neural networks. Neural Networks 24(8), 842–851 (2011)CrossRefGoogle Scholar
  12. 12.
    Chen, H., Carlsson, L., Eriksson, M., Varkonyi, P., Norinder, U., Nilsson, I.: Beyond the scope of free-wilson analysis: Building interpretable qsar models with machine learning algorithms. Journal of Chemical Information and Modeling 53, 1324–1336 (2013)CrossRefGoogle Scholar
  13. 13.
    Faulon, J.-L., Collins, M.J., Carr, R.D.: The signature molecular descriptor. 4. Canonizing molecules using extended valence sequences. J. Chem. Inf. Comput. Sci. 44(2), 427–436 (2004)CrossRefGoogle Scholar
  14. 14.
    Politis, D.N., Romano, J.P., Wolf, M.: Subsampling. Springer, New York (1999)CrossRefzbMATHGoogle Scholar
  15. 15.
    Egloff, D., Leippold, M.: Quantile estimation with adaptive importance sampling. The Annals of Statistics 38(2), 1244–1278 (2010)MathSciNetCrossRefzbMATHGoogle Scholar

Copyright information

© IFIP International Federation for Information Processing 2014

Authors and Affiliations

  • Lars Carlsson
    • 1
  • Martin Eklund
    • 2
  • Ulf Norinder
    • 3
  1. 1.AstraZeneca Research and DevelopmentMölndalSweden
  2. 2.Department of SurgeryUniversity of California San Francisco (UCSF)San FranciscoUSA
  3. 3.H. Lundbeck A/SValbyDenmark

Personalised recommendations