Logistic regression classifier is a statistical model that uses the logistic function to model a categorical target variable.
In this model, the probabilities describing the possible outcomes of a single trial are modeled using a logistic function.
Logistic regression does not require a linear relationship between inputs and output variables. This is due to applying a nonlinear log transformation to the odds ratio. If the probability is greater than 0.5, the predictions will be classified as class 1. Otherwise, class 0 is assigned.
The solvers are the optimization models used to minimize the cost function.
It is possible to use the following solvers:
lbfgs, sag and newton-cg solvers converge faster for some high-dimensional data.
The lbfgs solver is recommended for use for small datasets because performance drops with large datasets.
sag and saga are faster than other solvers for large datasets when both the number of samples and the number of features is large.
Logistic Regression is an efficient algorithm for finding linear-separated surfaces. It can scale to high volumes of data, and it can perform relatively well even with a reduced number of training samples. It natively provides outputs as probabilities.
The hyperparameters for this model type are:
- Optimization problem algorithm
- Inverse of regularization strength
- Stop condition tolerance
- Class weight