TradingView
RicardoSantos
22 janv. 2024 11:04

TimeSeriesClassificationActivationFunctions 

Bitcoin / U.S. dollarBitstamp

Description

Library "TimeSeriesClassificationActivationFunctions"
Provides some activation functions useful in time series classification.
___

reference:
github.com/PYFTS/pyFTS/blob/master/pyFTS/common/Activations.py

method scale(dist, weights)
  Activate values by a normalized scale.
  Namespace types: map<int, float>
  Parameters:
    dist (map<int, float>): Source distribution map.
    weights (map<int, float>): Weights distribution map.
  Returns: Normalized distribution map.

method softmax(dist, weights)
  Activate values with a softmax algorithm.
  Namespace types: map<int, float>
  Parameters:
    dist (map<int, float>): Source distribution map.
    weights (map<int, float>): Weights distribution map.
  Returns: Normalized distribution map.

method argmax(dist, weights)
  Activate values with a argmax algorithm.
  Namespace types: map<int, float>
  Parameters:
    dist (map<int, float>): Source distribution map.
    weights (map<int, float>): Weights distribution map.
  Returns: first key of argmax value of the transformed distribution.
Commentaires
KioseffTrading
Excellent!
Plus