Feature Engineering

Simple Machine Learning Tool Kit package

This package contains the modules to simplify your code for your feature engineering processes.

It is part of the educational repositories (https://github.com/pandle/materials) to learn how to write stardard code and common uses of the TDD.

Package contents two classes to manage feature engineering.

>>> import smltk
>>> help(smltk)
>>> from smltk.feature_engineering import Indicator
>>> help(Indicator)

# license MIT # support https://github.com/bilardi/smltk/issues

Indicator

The class Indicator contains the tool kit to calculate the principal indicators.

Indicators Tool Kit

Indicator.get_dc_events

Compute all relevant Directional Change parameters

Indicator.get_dc_events_starts

Get only Directional Changes starts

Indicator.plot_dc

Plot image with directional changes

Detailed list

class smltk.feature_engineering.Indicator(params={})

The class Indicator contains the tool kit to calculate the principal indicators.

Arguments: params (dict) with the keys below
events (list[str]):

list of directional change events

timeseries (list[int|float]):

list of values, default None

Here’s an example:

>>> from smltk.feature_engineering import Indicator
>>> timeseries = numpy.array()
>>> indicator = Indicator()
>>> dc_events = indicator.get_dc_events(timeseries)
>>> print(dc_events)
array['upward dc', 'downward dc', ..]
get_dc_events(timeseries: array = None, threshold: float = 0.0001) list

Compute all relevant Directional Change parameters

Arguments:
timeseries (list[int|float]):

list of values

threshold (float):

default is 0.0001

Returns:

list of directional change events

get_dc_events_ends(events: list = None, timeseries: list = None) dict

Get only Directional Changes ends

Arguments:
events (list[str]):

list of directional change events

timeseries (list[int|float]):

list of values

Returns:

dictionary of boolean lists when each directional change events ends

get_dc_events_starts(events: list = None, timeseries: list = None) dict

Get only Directional Changes starts

Arguments:
events (list[str]):

list of directional change events

timeseries (list[int|float]):

list of values

Returns:

dictionary of boolean lists when each directional change events starts

plot_dc(params: dict = {}, return_ax=False)

Plot image with directional changes

Arguments: params (dict) with the keys below
dc_colors (dict):

key-value about each event-color of directional change

events (list[str]):

list of events names for time point

timeseries (list[float]):

list of values

timestamp (list[int|datetime]):

time point list

figsize (tuple):

default (10, 5)

title (str):

title of plot

x_axis_label (str):

label of x axis

y_axis_label (str):

label of y axis

Returns:

plot or its object