luisDVA

# exploring comments about loaded packages
library(bigrquery) # CRAN v1.4.1
library(dplyr) # CRAN v1.1.1
library(stringr) # CRAN v1.5.0
library(readr) # CRAN v2.1.4
library(tidyr) # CRAN v1.3.0
library(rlang) # CRAN v1.1.0
library(purrr) # CRAN v1.0.1

luisDVA

library(ggsvg)
library(ggplot2)

# the svg that I need to use 
svg_url <- "https://raw.githubusercontent.com/luisDVA/luisdva.github.io/master/assets/images/Cat-green.svg"
cat_svg <- paste(readLines(svg_url), collapse = "\n")
# toy data
dat <-
  expand.grid(y=1:2,x=seq_len(2))
dat$qual <- letters[1:4]  

luisDVA

# lista 1
dat1a <- tibble(id=c("A","B","C","D","E"),val=rnorm(5))
dat1b <- tibble(id=c("A","B","C","D","E"),val=rnorm(5))

dat1 <- list(dat1a,dat1b)
dat1

# lista dos
dat2a <- tibble(id=c("A","D","E"),label=rnorm(3))
dat2b_1 <- tibble(id=c("A","B","C","E"),label=rnorm(4))

luisDVA

## %######################################################%##
#                                                          #
####        Regex for data cleaning 1 - your turn        ####
#                                                          #
## %######################################################%##

# After running the code below: 

library(ggplot2)
library(dplyr)

luisDVA

##%######################################################%##
#                                                          #
####               Regex in R - Your Turn               ####
#                                                          #
##%######################################################%##


# Match the following regular expressions against the test vector below using `str_detect`.  
## Can you explain the matches?

luisDVA

## %######################################################%##
#                                                          #
####               Letter case - your turn               ####
#                                                          #
## %######################################################%##


# Import the Marine Protected Areas dataset (MPAS-your.csv)
# Summarize the number of Marine Protected Areas by country (Country full).

luisDVA

## %######################################################%##
#                                                          #
####            Compound values - your turn             ####
#                                                          #
## %######################################################%##

# Import the Marine Protected Areas dataset (MPAS-your.csv)
# Separate the country codes variable (ISO3 and UN scheme)
# Unnest the Reference variable
# > Keep an eye on the separators

luisDVA

## %######################################################%##
#                                                          #
####         Unusable variable names - your turn         ####
#                                                          #
## %######################################################%##

# - Import the Marine Protected Areas data (MPAS-your.csv)
# - Make the variable names usable by placing all header fragments in a single
#   header row
# - Clean the names for consistency

luisDVA

## %######################################################%##
#                                                          #
####               Whitespace - your turn               ####
#                                                          #
## %######################################################%##

# - Import the Marine Protected Areas data (MPAS-your.csv) from the previous lesson
# - check the Country variable for leading or trailing whitespace
# - Remove it if necessary.

luisDVA

## %######################################################%##
#                                                          #
####             Broken values - your turn              ####
#                                                          #
## %######################################################%##

# Load the raw Age of Empires units dataset from csv (aoe_raw.csv)
# Identify the broken values in both the 'Type' and 'Name' columns and unbreak them
# Clean up any separator-related issues arising from the 'unbreaking'