pengelbrecht

import numpy as np
from tensorflow import keras
from tensorflow.keras import layers
import time

start_time = time.time()

# Model / data parameters
num_classes = 10
input_shape = (28, 28, 1)

pengelbrecht

var MAX_AGENTS_TO_TEST = 200;

function determineAgents(callDuration, calls, periodLength, answerTarget, serviceLevelTarget) {
  if(calls === 0) return(0);
  for (var agents = 1; agents <= MAX_AGENTS_TO_TEST; agents++) {
    if(serviceLevel(callDuration, calls, periodLength, agents, answerTarget) >= serviceLevelTarget) break;
  }
  return(agents);
}  

pengelbrecht

library(lubridate) # needs >= v1.6.0.9 for correct handling of lubridate.week.start with floor_date
options(lubridate.week.start = 1) # Weeks start on Mondays, use 7 for Sundays
library(dplyr)
library(ggplot2)
library(scales)
library(glue)
source("../analytics/utils/db.R", local = FALSE)
query <- function(sql, db = mainframe_db) {return(collect(tbl(db, sql(sql))))}

period_diff <- function(start, end, unit = "month") {

pengelbrecht

select 
    product_segment as Segment,
    round(avg(revenue_then)) as `Average Revenue`,
    count(distinct company_id) as n,
    round(power(power(sum(revenue_now) / sum(revenue_then),  1/period_diff(date_format(month_now, "%Y%m"), date_format(month_then, "%Y%m"))), 12), 2) as `Annual Revenue Retention`, 
    round(1 - power(1 - sum(churned)/count(*),  1/period_diff(date_format(month_now, "%Y%m"), date_format(month_then, "%Y%m"))), 3) as `Monthly Churn`
from (
select old.company_id, product_segment, size_segment, channel_segment, old.revenue as revenue_then, new.revenue as revenue_now, new.revenue is null as churned, new.revenue_month as month_now, old.revenue_month as month_then
from (
select 

pengelbrecht

library(RPostgreSQL)
library(dplyr)
library(whisker)
source("config.R")

redshift <<- src_postgres(
  dbname   = redshift_database,
  host     = redshift_server,
  port     = redshift_port,
  user     = redshift_username,

pengelbrecht

library(rvest)
library(ggvis)

trim <- function (x) gsub("^\\s+|\\s+$", "", x)

to_number <- function(string) {
  string <- trim(string)
  len <- nchar(string)
  last_char <-substr(string, len, len)
  suppressWarnings(

pengelbrecht

require(RMySQL)
require(ggplot2)
require(scales)

myusername = "peter"
mypassword = "sekret"

system('ssh -f [email protected]  -L 3306:localhost:3306 -N -o ExitOnForwardFailure=yes')
con <- dbConnect(MySQL(),
                 user=myusername, password=mypassword,

pengelbrecht

// Global variables
var BASEURI = 'https://app.firmafon.dk/api/v1';
var OPENTIME, CLOSETIME;
var SS = SpreadsheetApp.getActiveSpreadsheet();

function readConfig() {
  OPENTIME = new Date(); 
  OPENTIME.setHours(SS.getRangeByName("OpenHour").getValue());
  OPENTIME.setMinutes(SS.getRangeByName("OpenMinute").getValue());
  OPENTIME.setSeconds(0);

pengelbrecht

// Global variables
var BASEURI = 'https://app.firmafon.dk/api/v1';
var OPENTIME, CLOSETIME;
var SS = SpreadsheetApp.getActiveSpreadsheet();

function readConfig() {
  OPENTIME = new Date(); 
  OPENTIME.setHours(SS.getRangeByName("OpenHour").getValue());
  OPENTIME.setMinutes(SS.getRangeByName("OpenMinute").getValue());
  OPENTIME.setSeconds(0);

pengelbrecht

# Formulas from http://www.mitan.co.uk/erlang/elgcspsh.htm

#input
n = 80 # number of calls
t = 8*60*60 # period length in seconds
d = 4.5*60 # call duration in seconds
M = 2 # number of agents
s = 20 # target SLA in seconds

# Output