Model Validation:Simple ways of validating predictive models

Поділитися
Вставка
  • Опубліковано 15 гру 2024

КОМЕНТАРІ • 13

  • @padmanandm
    @padmanandm 5 років тому +2

    What do you mean by saying that the training data does not have the target variable? Without target variable how are you going to run the regression. The only difference between training and hold out data is that hold out observations won’t be in the training data.

  • @emilianarizopatron4280
    @emilianarizopatron4280 6 років тому +1

    Very helpful,
    One question,
    How do I transform integer values from a risk score into beta coefficients for external validation?
    Model:
    score=var1 - var2 + 18*(var 3)
    for each score (range) a probability of daeth is given. ie: score 0-15=0%, score 51-59=3%, etc.
    I want to transform a a simple risk tool values to obtain the beta coefficient for each predictor so I can validate (external validation) the score with another dataset. I am missing the beta coefficients from the original model, I just have the prediction rule.
    Thank you

  • @benedictmashimbibengelectm3199
    @benedictmashimbibengelectm3199 3 роки тому

    Very nice video....
    But what do you mean when you say betas should be around 0.5 or 0.4 etc

  • @johnsquarez874
    @johnsquarez874 4 роки тому

    very well explained and very helpful

  • @sukanyagupta9963
    @sukanyagupta9963 8 років тому +2

    Nicely explained

  • @SushmaReddy-hu2hf
    @SushmaReddy-hu2hf Рік тому

    Hi need help in model validation can you please reply

    • @AnalyticsUniversity
      @AnalyticsUniversity  Рік тому

      please drop me an email on analyticsuniversity@gmail.com or whatsapp me on : +31 625521289

  • @saikiraraju.m.r
    @saikiraraju.m.r 4 роки тому

    Sir, I have developed a logistic regression model by taking 70% data and predicted data for 30% data. But how to validate the model.....means only via confusion matrix and Predicted AUC ? Or any other methods?

  • @vivekgupta1296
    @vivekgupta1296 7 років тому

    very well explained. would like to ask few questions on validation, please advise if i can email my queries. thanks

  • @rajeshgalla12
    @rajeshgalla12 5 років тому

    Too many ads