Feature Scale

Feature scale is a process that scales each feature along column. Feature Scale module supports min-max scale and standard scale.

1. min-max scale: this estimator scales and translates each feature individually such that it is in the given range on the training set, e.g. between min and max value of each feature.
2. standard scale: standardize features by removing the mean and scaling to unit variance

Param

scale_param

Attributes

Classes

ScaleParam(method='standard_scale', mode='normal', scale_col_indexes=-1, scale_names=None, feat_upper=None, feat_lower=None, with_mean=True, with_std=True, need_run=True)

Bases: BaseParam

Define the feature scale parameters.

Parameters:

Name	Type	Description	Default
method		like scale in sklearn, now it support "min_max_scale" and "standard_scale", and will support other scale method soon. Default standard_scale, which will do nothing for scale	"standard_scale"
mode		for mode is "normal", the feat_upper and feat_lower is the normal value like "10" or "3.1" and for "cap", feat_upper and feature_lower will between 0 and 1, which means the percentile of the column. Default "normal"	"normal"
feat_upper	int or float or list of int or float	the upper limit in the column. If use list, mode must be "normal", and list length should equal to the number of features to scale. If the scaled value is larger than feat_upper, it will be set to feat_upper	None
feat_lower		the lower limit in the column. If use list, mode must be "normal", and list length should equal to the number of features to scale. If the scaled value is less than feat_lower, it will be set to feat_lower	None
scale_col_indexes		the idx of column in scale_column_idx will be scaled, while the idx of column is not in, it will not be scaled.	-1
scale_names	list of string	Specify which columns need to scaled. Each element in the list represent for a column name in header. default: []	None
with_mean	bool	used for "standard_scale". Default True.	True
with_std	bool	used for "standard_scale". Default True. The standard scale of column x is calculated as : $z = (x - u) / s$ , where $u$ is the mean of the column and $s$ is the standard deviation of the column. if with_mean is False, $u$ will be 0, and if with_std is False, $s$ will be 1.	True
need_run	bool	Indicate if this module needed to be run, default True	True

Source code in python/federatedml/param/scale_param.py

def __init__(
        self,
        method="standard_scale",
        mode="normal",
        scale_col_indexes=-1,
        scale_names=None,
        feat_upper=None,
        feat_lower=None,
        with_mean=True,
        with_std=True,
        need_run=True):
    super().__init__()
    self.scale_names = [] if scale_names is None else scale_names

    self.method = method
    self.mode = mode
    self.feat_upper = feat_upper
    # LOGGER.debug("self.feat_upper:{}, type:{}".format(self.feat_upper, type(self.feat_upper)))
    self.feat_lower = feat_lower
    self.scale_col_indexes = scale_col_indexes

    self.with_mean = with_mean
    self.with_std = with_std

    self.need_run = need_run

Attributes

scale_names = [] if scale_names is None else scale_names instance-attribute

method = method instance-attribute

mode = mode instance-attribute

feat_upper = feat_upper instance-attribute

feat_lower = feat_lower instance-attribute

scale_col_indexes = scale_col_indexes instance-attribute

with_mean = with_mean instance-attribute

with_std = with_std instance-attribute

need_run = need_run instance-attribute

Functions

check()

Source code in python/federatedml/param/scale_param.py

def check(self):
    if self.method is not None:
        descr = "scale param's method"
        self.method = self.check_and_change_lower(self.method,
                                                  [consts.MINMAXSCALE, consts.STANDARDSCALE],
                                                  descr)

    descr = "scale param's mode"
    self.mode = self.check_and_change_lower(self.mode,
                                            [consts.NORMAL, consts.CAP],
                                            descr)
    # LOGGER.debug("self.feat_upper:{}, type:{}".format(self.feat_upper, type(self.feat_upper)))
    # if type(self.feat_upper).__name__ not in ["float", "int"]:
    #     raise ValueError("scale param's feat_upper {} not supported, should be float or int".format(
    #         self.feat_upper))

    if self.scale_col_indexes != -1 and not isinstance(self.scale_col_indexes, list):
        raise ValueError("scale_col_indexes is should be -1 or a list")

    if self.scale_names is None:
        self.scale_names = []
    if not isinstance(self.scale_names, list):
        raise ValueError("scale_names is should be a list of string")
    else:
        for e in self.scale_names:
            if not isinstance(e, str):
                raise ValueError("scale_names is should be a list of string")

    self.check_boolean(self.with_mean, "scale_param with_mean")
    self.check_boolean(self.with_std, "scale_param with_std")
    self.check_boolean(self.need_run, "scale_param need_run")

    LOGGER.debug("Finish scale parameter check!")
    return True

---

最后更新: 2021-11-15