# Logistic regression using a probability as Input

I am creating a model that predicts the availability of machines. Historical data is given as a percentage of time that the machine is available. (e.g. machine x was available 95% of the time)

I would thus compute a logit specification model with the dependant variable not being 0 or 1, but within the interval (0,1). Is it valid and possible to use logistic regression on this scenario?

• Hi Machine, welcome! Short answer: yes.
– Jim
May 11, 2018 at 10:09

The answer depends on what exactly is your data. If your data are sample sizes $n_i$ and observed proportions of success $p_i$, then you can use logistic regression to model such data. On another hand, if you only have values in $(0, 1)$ that you want to model, then this kind of data is to be modeled with beta regression, a regression model for conditional distribution that is bounded on both sides, modeled using beta distribution parametrized by mean and precision.