aflansburg

# import time - not the abstract construct of 'time'
# but rather a library built into Python for
# dealing with time
from time import time

# ML stuff
ada_tuned_clf = AdaBoostClassifier(random_state=1)

# some canned params for hypertuning
parameters = {

aflansburg

# runtime info based on solution below and fit_time results of the gridsearchcv return object
# based on a response on StackExchange Data Science - Naveen Vuppula
# https://datascience.stackexchange.com/a/93524/41883
# from time import time
def gridsearch_runtime(grid_obj, X_train, y_train):
  '''
  Parameters:
    grid_obj: GridSearchCV return object that has not yet been fit to training data
    X_train: split training data independent variables
    y_train: split training data containing dependent variable

aflansburg

# import libs
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt

# this function will create a boxplot + histogram plot using Seaborn's Jointgrid
# we'll also provide Type annotations to provide hints to future users
def dual_plot(series: pd.Series, figsize: tuple = (16,8),
              bins: int = None, return_plot: bool = False,

aflansburg

# function to iterate over specified variables and view their value counts

# add typing to help understand our function if reused elsewhere
from typing import List
def value_count_rep(columns: List, df: pd.DataFrame) -> None:
    '''
    Parameters: List of columns to iterate over
    Returns:    No return value. Prints the value counts of each column(feature) to stdout
    '''
    for column in columns:

aflansburg

# https://www.python.org/dev/peps/pep-0634/
# PEP 634 proposed (and was accepted) Structural Pattern Matching (switch/case)
# for Python (available in 3.10) - as of this Gist,
# prerelease 3.10.0b2 is available

import inspect

F = [
  lambda x,y: x+y,
  lambda x: x+1,

aflansburg

## handy multi-plot function for showing mode, median, and mean lines in a distplot 
## (but using histplot since distplot is deprecated)
## Author - Abram Flansburg
## Intended for use in Jupyter / iPython
def skewness_plot(series):
    """
    Plots a series using the histogram plot with kde and plots, mean, mode, and median lines.
    *** Dependencies ***
    Series must be a pandas.Series
    Seaborn must be imported as sns

aflansburg

// In this scenario - a Cloudwatch Subscription Filter is hooked up to a Node.js lambda's ARN 
// (see AWS docs around Log Group subscription filters)
// Will need a webhook for the appropriate Slack channel

const https = require('https');
const zlib = require('zlib');

const options = {
    hostname: 'hooks.slack.com',
    path: 'SLACK_WEBHOOK_URL',

aflansburg

# Not my original function - looking for citation
def missing_check(df):
    null_val_sum = df.isnull().sum()
    total = df.isnull().sum().sort_values(ascending=False) # total null values
    percent = (null_val_sum/df.isnull().count()).sort_values(ascending=False)
    missing_data = pd.concat([total, percent], axis=1, keys=['Total', 'Percent'])
    return missing_data
'''
Example:
missing_data(df_with_nulls)

aflansburg

# imagine this lives at app/lib/complicated_data
module ComplicatedData
  def generate_complicated_plot_data(start_date: 30.days.ago, end_date: Date.today, type: nil)
    # implementation
  end
  
  # module_function ensures the method cannot be overridden or extended
  module_function :generate_complicated_plot_data
end

aflansburg

# radius of earth in meters
R = 6371000;

def haversine(coord1:, coord2:)
  # first add our own radians function since Ruby Math does not have one
  radians = -> (degrees) { degrees * (Math::PI / 180)}

  # convert latitude degrees to radians
  phi_1 = radians.call(coord1[:latitude]);
  phi_2 = radians.call(coord2[:latitude]);