Multiple correspondence analysis

Поділитися
Вставка
  • Опубліковано 29 січ 2025

КОМЕНТАРІ • 5

  • @sinan_islam
    @sinan_islam 8 місяців тому +2

    you need to create playlist for multiple correspondence analysis

  • @ukelelelogy
    @ukelelelogy Рік тому +2

    How do you know what each dimension represents?

    • @statisticsninja
      @statisticsninja  Рік тому +3

      Hey Karla,
      Interpreting dimensions is more an art than a science. I take a column principal component, (the colpcoord slot of ca::mjca output) and then look at which variables have large positive and which variables have large negative values. Then I make a judgment call on as to why the variables are grouped that way. I ignore variables with small values.

  • @emmaquazi8815
    @emmaquazi8815 Рік тому +1

    Thus was very good explanation. My question how do i was the results of MCA to do cluster analysis.

    • @statisticsninja
      @statisticsninja  Рік тому

      You can cluster the rows of data using Row principal coordinates (rowpcoord slot in ca::mjca() output), you can cluster the variable combinations using Column principal coordinates (colpcoord slot in ca::mjca() output).