+ Reply to Thread
Results 1 to 4 of 4

Thread: Logistic regression : negative coefficients, categorical variables

  1. #1
    Points: 8, Level: 1
    Level completed: 15%, Points required for next Level: 42

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Logistic regression : negative coefficients, categorical variables




    Hi
    I am trying to build attrition model and have these variables :- 10 categorical variables with sub levels which created into dummy variables and 4 continuous variables in a trained data set. My Y dependent variable is binary variable. After running log regress, the results are some of the dummy variables are found to have collinearity error and some of them are insignificant. Tried running penalized log reg, same results obtained. After omitting the dummy variables, I run a new log regress, there are negative coefficients in the 3 continuous variables which are found significant. But, they are in a very small numbers e.g. -0.003,-0.005 etc which did not make senses. Now I am in a fix.

    Did i go wrong somewhere as all the categorical variables are not significant?
    Or I should not be using logistic regress?

    All advises appreciated!

  2. #2
    Omega Contributor
    Points: 38,374, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,998
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: Logistic regression : negative coefficients, categorical variables

    Do you have a huge dataset?
    Stop cowardice, ban guns!

  3. #3
    Points: 8, Level: 1
    Level completed: 15%, Points required for next Level: 42

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Logistic regression : negative coefficients, categorical variables

    In total I have 107 rows. Not big actually.

  4. #4
    TS Contributor
    Points: 12,227, Level: 72
    Level completed: 45%, Points required for next Level: 223
    rogojel's Avatar
    Location
    I work in Europe, live in Hungary
    Posts
    1,470
    Thanks
    160
    Thanked 332 Times in 312 Posts

    Re: Logistic regression : negative coefficients, categorical variables


    hi,
    without knowing the details, you could look ar two aspects: the odds ratio will depend on the exponential of the coefficient, and it will give you the effect of increasing the continuous variable by 1 unit. If the unit is small, the odds ratio will likely be very close to 1, and you would need to calculate the odds ratio for some practically significant change in the continuous variable.

    E.g. odds of heart attack vs. weight. If you measure the weight in grams, the odds will not greately change by a weight change of 1 gram, though they may change a lot by an increase of say, 5 kg.

    regards

  5. The Following User Says Thank You to rogojel For This Useful Post:

    gianmarco (04-22-2017)

+ Reply to Thread

           




Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats