--- title: "Lending Club Data" author: "Eric A. Suess" date: "2/16/2021" output: html_document --- ```{r setup, include=FALSE} knitr::opts_chunk$set(echo = TRUE) ``` ## Lending Club Data ```{r} library(pacman) p_load(tidyverse, fs, tictoc, lubridate, beepr) ``` ## ```{r} tic() lending_club_data <- read_csv("data_accepted/accepted_2007_to_2018Q4.csv") toc() ``` ```{r} head(lending_club_data) str(lending_club_data) ``` ```{r} # The issue_d variable will be used to determine year the loan was issued lending_club_data <- lending_club_data %>% mutate(year = year(mdy(issue_d))) # check the unique years in the data lending_club_data %>% select(year) %>% distinct() %>% arrange(year) # subset the years 2012-2014 for the training data lending_club_data_2012_2014 <- lending_club_data %>% filter(year %in% c(2012, 2013, 2014)) # check the unique years in the data lending_club_data_2012_2014 %>% select(year) %>% distinct(year) %>% arrange(year) # subset the year 2015 for the testing data lending_club_data_2015 <- lending_club_data %>% filter(year == 2015) # check the unique year in the data lending_club_data_2015 %>% select(year) %>% distinct(year) %>% arrange(year) ``` ```{r} lending_club_data_2012_2014 %>% group_by(year) %>% count(year) lending_club_data_2015 %>% count(year) ``` ## ```{r} if (!is_dir("data")) dir_create("data") is_dir("data") ``` ```{r} tic() write_csv(lending_club_data_2012_2014, "data/lending_club_data_2012_2014.csv") toc() tic() write_csv(lending_club_data_2015, "data/lending_club_data_2015.csv") toc() ``` ```{r} tic() saveRDS(lending_club_data_2012_2014, "data/lending_club_data_2012_2014.Rds") toc() tic() saveRDS(lending_club_data_2015, "data/lending_club_data_2015.Rds") toc() ``` ```{r} file_show("data") ``` ## Reading ```{r} tic() lending_club_data_2012_2014_new <- read_csv("data/lending_club_data_2012_2014.csv") toc() tic() lending_club_data_2015_new <- read_csv("data/lending_club_data_2015.csv") toc() ``` ```{r} tic() lending_club_data_2012_2014_new <- readRDS("data/lending_club_data_2012_2014.Rds") toc() tic() lending_club_data_2015_new <- readRDS("data/lending_club_data_2015.Rds") toc() ``` ```{r} beep();beep();beep();beep(sound=4) ```