madagra

Strategy	Cross-validation set	Test set	Forecast
Direct	14.302	8.378	13.124
Recursive	14.302	8.378	16.471

madagra

import re
from anytree.search import findall

def find_by_keyword(self, keyword) -> Node:
    """
    Find the node whose regex matches the keyword given as input
    """
    def match_regex(node):
        has_regex = hasattr(node, "regex") and node.regex is not None
        if has_regex:

madagra

from anytree.exporter import DotExporter

COLOR_SCHEME = ["aliceblue", "antiquewhite", "azure", "coral", "palegreen"]
    
def export(self):
    """
    Export tree into graphviz format in both *.dot file and image *.png file
    """

    def nodeattr_fn(node):

madagra

import json
from anytree import Node, RenderTree
from anytree.search import find

class TaxonomyParser:

    """
    This class is a wrapper on a Tree class from the anytree library to hold
    different data hierarchies read from a JSON representation
    """

madagra

{
  "L0": [
    { "name": "device" }
  ],
  "L1": [
    { "name": "Networking", "parent": "Device" },
    { "name": "Computer", "parent": "Device" }
  ],
  "L2": [
    { "name": "Router", "parent": "Networking" },

madagra

{
  "name": "Device", 
  "children": [ 
    {
      "name": "Networking", 
      "children": [
         { "name": "Router", "children": [] },
         { "name": "Switch" },
         { "name": "Firewall", "children": [] }
      ]

madagra

app = dash.Dash(__name__)

app.layout = html.Div(children=[
    html.Div(dcc.Markdown(description)),
    html.Hr(),
    html.Div([
        html.Div(id="parameters-box-id", children=[
            dcc.Markdown(params_description),
            dcc.Input(
                id="tottime-input-id",

madagra

def direct_forecast(y, model, lags, 
                    n_steps=FCAST_STEPS, step="1H"):
    
    """    
    Parameters
    ----------
    y: pd.Series holding the input time-series to forecast
    model: a ML model not trained
    lags: list of lags used for training the model
    n_steps: how many steps forecast into the future

madagra

def recursive_forecast(y, model, lags, 
                       n_steps=FCAST_STEPS, step="1H"):
    
    """
    Parameters
    ----------
    y: pd.Series holding the input time-series to forecast
    model: pre-trained machine learning model
    lags: list of lags used for training the model
    n_steps: number of time periods in the forecasting horizon

madagra

import pandas as pd
from sklearn.preprocessing import StandardScaler
from statsmodels.tsa.stattools import pacf

def create_lag_features(y):
   
    scaler = StandardScaler()
    features = pd.DataFrame()
    
    partial = pd.Series(data=pacf(y, nlags=48))