wesslen

library(pwr)

# Cohen suggests that f values of 0.1, 0.25, and 0.4 represent small, medium, and large effect sizes respectively.

f <- seq(0.05, 0.5, 0.05)

pwrANOVA <- function(i){
  z <- pwr.anova.test(k = 3, n = NULL, f = i, sig.level = 0.05, power = 0.8)
  
  as.data.frame(t(unlist(z)))

wesslen

import glob
import json
from pymongo import MongoClient

# fill in hostname and port
HOST = "datalake2.local" #hostname
PORT = 27017

# connect to client
client = MongoClient(HOST, PORT)

wesslen

library(jsonlite)

dir <- "./data/" # files in this folder

files <- list.files(dir)
files <- paste0(dir,files[grepl("Flume",files)])

readIn <- function(fileName){
  stream_in(file(fileName,open="r"))
}

wesslen

library(tidyverse)

path <-"~/Dropbox_UNCC/Dropbox (UNC Charlotte)/xai-cfpb/data/Consumer_Complaints.csv"
Consumer_Complaints <- read_csv(path) %>%
  rename(text = `Consumer complaint narrative`) %>%
  filter(!is.na(text)) %>%
  mutate(Bank = case_when(
    str_detect(Company, "BANK OF") ~ "BAC",
    str_detect(Company, "CITIBANK") ~ "CITI",
    str_detect(Company, "JPMORGAN") ~ "JPM",

wesslen

#turn off GPU if needed
#Sys.setenv("CUDA_VISIBLE_DEVICES" = -1)  

library(telegram.bot); library(keras); library(lime); library(magick); library(httr); library(tidyverse)

# keras pre-trained model vgg16
model <- application_vgg16(weights = "imagenet",include_top = TRUE)

# make sure to set up auth via .environment 
# see https://github.com/ebeneditos/telegram.bot/wiki/Tutorial-%E2%80%93-Building-an-R-Bot-in-3-steps

wesslen

# make sure you have tidyverse installed
# install it using: "install.packages('tidyverse')"
library(tidyverse)

# set parameters
n_part = 10 # number of participants per group
n_task = 5 # number of tasks

low =
  data_frame(

wesslen

## load ggplot2
library(ggplot2)

## function to make basic hex sticker
make_hexsticker <- function(pkg, pkg_size, pkg_color, pkg_y = -.33,
                            bg, border,
                            icon, icon_color, icon_size, family, face = "plain",
                            url, url_size = 8, url_x = .225, url_y = -.76,
                            url_color = "transparent",
                            fa = "Font Awesome 5 Brands",

wesslen

library(rtweet)

# make sure to set up auth, see https://rtweet.info/index.html

rt <- stream_tweets(
  q = c(-125, 26, -65, 49), # united states
  timeout = 30, # run for 30 seconds; set to Inf to run indefinitely
  file_name = NULL, # name of file name e.g., paste0("stream",gsub(" |-|:", "", Sys.time()),".json")
  dir = NULL # directory to put files into
)

wesslen

# from https://rdrr.io/github/thomasp85/lime/src/R/plot_image.R

tidy_raster <- function(im) {
  if (!requireNamespace('magick', quietly = TRUE)) {
    stop('The magick package is required for image explanation', call. = FALSE)
  }
  raster <- as.raster(im)
  data.frame(x = rep(seq_len(ncol(raster)), nrow(raster)),
             y = rep(seq_len(nrow(raster)), each = ncol(raster)),
             colour = as.vector(raster),

wesslen

# see https://arxiv.org/abs/1709.03856
# see http://www.bnosac.be/index.php/blog/86-neural-text-modelling-with-r-package-ruimtehol

library(ruimtehol)
library(tidyverse)
df <- read_csv("complaints-2019-03-30_10_53.csv", 
               col_types = cols(`Complaint ID` = col_character(), 
                                `Date received` = col_character())) %>%
  rename(text = `Consumer complaint narrative`)