Bike Co: Text Mining and Cleaning

# Bike Shop Text Analysis 

library(tidyverse)
library(lubridate)

# Read in Data
bike_orderlines_tbl <- read_rds("C:/Users/open/Documents/R PROJECTS/00_data/bike_sales/data_wrangled/bike_orderlines.rds")

bike_orderlines_tbl

bikes_tbl <- readxl::read_excel("C:/Users/open/Documents/R PROJECTS/00_data/bike_sales/data_raw/bikes.xlsx")

bikes_tbl

# Cleaning and Organizing Data into Usable Column Data 
bikes_cleaned_tbl <- bikes_tbl %>% 
    select(model) %>% 
    mutate(model = case_when(
        model  == "CAAD Disc Ultegra" ~ "CAAD12 Disc Ultegra",
        model == "Syapse Carbon Tiagra" ~ "Synapse Carbon Tiagra",
        model == "Utegra" ~ "Supersix Evo Hi-Mod Utegra",
                             TRUE ~ model)) %>% 
    
    separate(col   = model, 
             into  = str_c("model_", 1:7), 
             sep   = " ", remove = FALSE,
             fill  = "right", 
             extra = "drop") %>%
    
    mutate(model_base = case_when(
        str_detect(str_to_lower(model_1), "supersix") ~ str_c(model_1, model_2, sep = " "),
        str_detect(str_to_lower(model_1), "fat") ~ str_c(model_1, model_2, sep = " "),
        str_detect(str_to_lower(model_1), "beast") ~ str_c(model_1, model_2, model_3, model_4, sep = " "),
        str_detect(str_to_lower(model_1), "bad") ~ str_c(model_1, model_2, sep = " "),
        str_detect(str_to_lower(model_2), "29") ~ str_c(model_1, model_2, sep = " "),
            TRUE ~ model_1)) %>% 
    
    mutate(model_tier = model %>% 
        str_replace(model_base, replacement = "") %>% str_trim()
    ) %>% 
    select(-matches("[0-9]")) %>% 
    
    mutate(
        black    = model_tier %>% str_to_lower() %>% str_detect("black") %>% as.numeric(),
        hi_mod   = model_tier %>% str_to_lower() %>% str_detect("hi_mod") %>% as.numeric(),
        team     = model_tier %>% str_to_lower() %>% str_detect("team") %>% as.numeric(),
        red      = model_tier %>% str_to_lower() %>% str_detect("red") %>% as.numeric(),
        ultegra  = model_tier %>% str_to_lower() %>% str_detect("ultegra") %>% as.numeric(),
        dura_ace = model_tier %>% str_to_lower() %>% str_detect("dura ace") %>% as.numeric(),
        disc     = model_tier %>% str_to_lower() %>% str_detect("disc") %>% as.numeric()
        )

View(bikes_cleaned_tbl)