# Scaling predictors while predicting dichotomous outcome

I have five predictors $x_1, x_2, x_3, x_4, x_5$ and a single binary response variable $y(0,1)$ in my dataset. All my predictors are continuous variables.

A preliminary exploratory analysis of my predictor variables show quite a few outliers in each of these predictors, as shown in the image below

How can i normalize these values within each of these 5 predictors such that the outliers do not significantly impact my prediction ? I am not in favour of getting rid of those observations that contain outliers because I am dealing with a sparse dataset so i like to use as much information as possible. Any advise or suggestion is much appreciated.