Distribution Optimizer

class distribution_optimizer

Pyhon software coming soon to GitHub

Transform raw data to a gaussian distributed space using a performance metric. If the data has been found to be gaussian distributed it can be used to:

  • Score feature performance in multi-classification models

  • Dimensionality reduction via gaussian processes

Augment Feature

Optimize Feature

Distribution Analysis

Signal Processing

Calculate Performance

fitting_method

The fitting method is a parameter which determines how to narrow down the potential transformation applied to the raw data. There are several to choose from and can greatly change the run time of your program. ​​

  • optimize     - Checks skew and kurtosis to qualify raw data transformations

  • robust         - Applies every possible transformation

  • magnitude - Rejects standard log transformation if any data is negative

skew_left

The skew left is the threshold for a negatively skewed distribution and therefore should be a negative number. The parameter is only relevant when the property fitting_method is set to optimize. The default value of the property is set to -1.

skew_right

The skew right is the threshold for a positively skewed distribution and therefore should be a positive number. The parameter is only relevant when the property fitting_method is set to optimize. The default value of the property is set to 1.

© 2018 by Alex Geiger

  • Facebook Social Icon
  • LinkedIn Social Icon
  • Google+ Social Icon