15 Matching Annotations
  1. Last 7 days
    1. Cross validation

      Should this be applied to the GBM approach? Would it prevent it collapsing onto a spike at 0.42?

    2. propensity scores is disc

      so how does it do on NSW?

    3. overlap between groups

      overlap of what?

    4. central region of true probabilities

      not sure what this means in practice

    5. effect

      "affect"

    6. completed in

      "carried out"

    7. propensity scores f

      perhaps good to state in text what would be the ideal distribution of propensity scores and which one of these propensity score estimations would be most useful

    8. ility of 0.42%.

      42%

    9. prevalent

      "evident" might be a better word?

    10. baggin

      bagging

    11. important

      importance

    12. s calculated as:

      I've often wondered whether probabilities can be averaged over all trees not just the indicator of class prediction

    13. ferred to as

      mtry is an argument specific to the R package randomForest, I imagine...

    14. mtry

      This is an argument for the R package randomForest -- only specific to this package I imagine?

    15. Probability prediction is not a typical machine learning task

      I wonder whether this is true -- as you say probabilistic classification is a thing. Isn't that the same?