Z-score

Also known as standard score of an observation, this method assumes data follows a gaussian distribution.

It's a parametric method which indicates how many standard deviations an instance is from the sample’s mean.

The z-score of every data point is calculated using the formula: $z = (x-\mu)/\sigma$ . It can be easily calculated using the method provided by sklearn.

PreviousIQR NextF-statistic

Last updated 5 years ago

Also known as standard score of an observation, this method assumes data follows a gaussian distribution.

It's a parametric method which indicates how many standard deviations an instance is from the sample’s mean.

The z-score of every data point is calculated using the formula: $z = (x-\mu)/\sigma$ . It can be easily calculated using the method provided by sklearn.

Once every z-score is computed, outliers are detected given a threshold. It's usually set to: $2.5$ , $3.0$ or $3.5$ .