learningR.Rmd

---
title: "learningR"
author: "Alejandro Hagan"
date: "2022-08-27"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```
#Working directory and file manipulation concepts

- Before you start working you need to set up the space you are working in. 
- If you don't have a git or version sharing repository
- When you save a file, data file, read inputs or otherwise, the folderpath branching will generally speaking come from here

- `getwd()` will tell you what is your current working directory
- `setwd()` can be used to change your working directly

Other popular system funtions are

- the abilily to list files or filders in a folder path
list.files(pattern=$^, recursive=T)

-the ability to check if a file exists?

file.exists dir.exists basename("c:/filedir/file.txt") -> "file.txt dirname("c:/filedir/file.txt")->

- create files
create.dir("./newdir") file.create("filename.txt") shell.exec("filename.txt")

filesstrings::move_file("c:home/filename.txt","c:newdir") file.copy("filename", "destination") file.rename(directory,"existing file name","new file name") -if directory is blank will assume getwd() file.choose() base::unlink("filename", recursive=T) -recursve must be T to delete directories

Link to learn more

how to connect to github

while all this may seem to new, quite honestly its because we have never been told good versioning and collaboration practices so we are reliant on saving to a LAN system, amending the name with versioning control

create a new github repsository in github

go to global settings (not repository settings), go to develrop settings and click personal access tokens

generate new token, copy (consider saving as you can only see it once)

usethis::git_branch_default()

check local branch

gert::git_remote_ls()

get information

resources

{r}
usethis::use_git_remote("origin",url=NULL,overwrite=TRUE)
usethis::use_github()
usethis::gh_token_help()

usethis::use_git_remote(name = "origin",url = "https://github.com/alejandrohagan/learningR.git",overwrite=TRUE)

gitcreds::gitcreds_set(url = "https://github.com/alejandrohagan/learningR.git")

usethis::use_github()
usethis::git_default_branch()

gh::gh_whoami()

usethis::git_remotes()

usethis::pr_push()

usethis::pr_fetch()

usethis::pr_pull()

usethis::pr_init(branch = "main")



usethis::use_git_remote(name = "origin",url = "https://github.com/alejandrohagan/learningR.git",overwrite=TRUE))

• To create a personal access token, call `create_github_token()`
• To store a token for current and future use, call `gitcreds::gitcreds_set()`

5 steps to change GitHub default branch from master to main | R-bloggers

Don't Lose your HEAD over Default Branches | R-bloggers

Git: Moving from Master to Main | R-bloggers

push & committ and commenting

How to manipulate & tidy data

how to import data

read.csv()

read_csv()

fread()

how to change column types

how to create data

seq() runif() sample(data,# of times, replace,) c() data.frame()

how to create data with basic loops / repitition

types of data

vector data.frame list

some base R basics that will be helpful to you as you read the forums

how to import files

by folder

by website

excel spreadsheet

powerbi model

how to find files by their type

how to append files together

how to automate file importation

file column names

how to change column names

statically

dynamically

best practices when naming columns

how to change column types

statically

dynamically

how to check data structure

-unique

how to clean data

how to shape data

how to subset data

filter

dynamic

static

select

dynamic

static

grouby

summarize

mutate

fill / blanks/ replace values

recode

pivot / unpivot

merge

iterate

apply a single function (single input or multiple inputs) to every/ some/ based on criteria column and get its outputs into a data frame (to be accessed)

apply a single function with multiple inputs

apply a functions to nested data frames

explatory analysis

start with variance analysis

how to extract information out of a table

for (i in range){

do something }

basic framework 1. is to define the vector that you want as an output and its size with vector("type",vectorsize) 2. assign that vector names with names() so that data has headings 3. create for (i in range) 4. define the task it will do individually to thedataset 4. assign that to the output vector

tips & tricks

check the task individually against a single element to ensure it works

you can extract other elements (typically row / column names) using other techniques (names()) and assign to output (doesn't need to be in the loop) -use seq_along(), ncol(), [[]] to define parameters and extract elements

{r loop_example, message=FALSE, warning=FALSE}
output <- vector("double",ncol(mtcars)) #assigns the output vector

cols_names <- names(mtcars) # assigns names of the dataframe to variable
names(output) <- cols_names # assigns the names 


for (i in cols_names) { # sometimes useful to use seq_along() here as well
  
  
  output[i] <- mean(mtcars[[i]]) # the double brackets ensures we only take out one item, this needs to be adjusted if two iems are expected, change the vector to a list
  
}

output


### Categorical Variables

In general you will need to distinguish betwen your character values as either straight character or categorical with levels

This becomes critical as you look to create categories and relationships in your data

forecats is the gotopackage in particular:

rename

recode() to change values in column

recode(col,newvalue=oldvalue)

reorder

fct_relevel

fct_relevel(col,level1,level2,etc)

fct_reorder(col,col_to_be_reorderby,function)

3)group variables into another group

case_when()

-typically used in combination with mutate() -can reference multiple conditions

case_when(col1==var1 ~ val1,             col1==var2 & col3==var3 ~ var 2,             is.na(col1) ~ "missingvalue",             TRUE ~ "defaultvalue )` cut()

##purr

{ } is used to stop a data frame from passing into as first agurmen

. is a place holder for the data frame

{r }
list.len=3
str(mpg,list.len=3)

str(mpg)
listviewer::jsonedit(mpg)

#patchwork

can organize with easy convention +,/,| is two charts on top and one chart beneath

but can also supplement with additional functions

plot_layout can also arrnage by rows

plot_layout(nrow = 3, byrow = FALSE) arguments: width= changes the graphs relative width size, when given as a numeric c(2,1) then the first columsn graphs are twice as large as the second columns height= changes the graphs reltive row heigh, ncol= numeric, changes number of columsn guides="collect" to remove duplicate guides theme(legend.position='bottom') moves the legend position

guide_area() to create area that guides=collect move towards

https://patchwork.data-imaginist.com/reference/plot_layout.html

plot_annotation() to add annotation title = 'The surprising story about mtcars' tag_levels = 'I' or "A" or "1" to set tag on each plot caption="Text" theme = theme(plot.title = element_text(size = 16))

use the below to add a blank text tile next to a plot grid::textGrob('Some really important text') or a table gridExtra::tableGrob(mtcars$$1:10, c('mpg', 'disp')$$)

plot_spacer() inserts an empty plot

inset_element() to insert a sub graph ontop of a new one

left = 0.6, bottom = 0.6, right = 1, top = 1 align_to = 'full

helpful tips:

When creating a patchwork, the resulting object remain a ggplot object referencing the last added plot. This means that you can continue to add objects such as geoms, scales, etc. to it as you would a normal ggplot:When creating a patchwork, the resulting object remain a ggplot object referencing the last added plot. This means that you can continue to add objects such as geoms, scales, etc. to it as you would a normal ggplot:geom_jitter(aes(gear, disp))

Often, especially when it comes to theming, you want to modify everything at once. patchwork provides two additional operators that facilitates this. & will add the element to all subplots in the patchwork, and * will add the element to all the subplots in the current nesting level. As with | and /, be aware that operator precedence must be kept in mind.

str_replace_all //s+ = all white spaces

how to write tables

font 1. Numerical data is right-aligned 2. Textual data is left-aligned 3. Headers are aligned with their data 3½. Don't use center alignment.

#visual guide ## axis title - axis title always all caps - align top y axis or left axis - color to match axis color ## graph title left alignment

Across(), if_any,if_all

summarize/mutuate/pivot_longer/pivot_wider

across // character based

starts_with

ends_with

contains

matches

num_range()

last_col

where()// with a function that has bolean condition eg. is.numeric

used to select columns by name, position or type (requires where() wrapp)

c(column names), position or type (where)

function, or list( function1=function(), function2=function())

{} is used to refenence preivously declared variables in the glue package or in functions that rerence glue package

some attributes have sepcial references, such as {.col} to reference a column and {.fn} to refernece a function

used in the .names argument of across

`across()` doesn't work with `select()` or `rename()`

mutate, group_by,count,distinct,summarize

filter is excluded and instead use if_any and if_all with exceltiion of

filter(across(everything), ~function)

Examples for filter

* `if_any()` keeps the rows where the predicate is true for *at least one* selected

column:

```{r}

starwars %>%

filter(if_any(everything(), ~ !is.na(.x)))

```

* `if_all()` keeps the rows where the predicate is true for *all* selected columns:

```{r}

starwars %>%

filter(if_all(everything(), ~ !is.na(.x)))

```

* Find all rows where no variable has missing values:

```{r}

starwars %>% filter(across(everything(), ~ !is.na(.x)))

```

Need to investigate rename_with and and itsimpact on select as it appears to be superseded

dplyr/colwise.Rmd at main · tidyverse/dplyr · GitHub

glamour of graphics

alignemtn

top left aligned to the chart left (plot.titile.position="plot"

add_count(dim,name="text") %\>% mutate(colname= glue::glue("{col}{text}")

rotate lebels, either by swapping axis or removing axis all together

remove borders

remove gridlines

left /right align text to create clean borders

indicate legend in title

graphing tips and tricks

if you want to plot a subset of the data but show atrend agains the full data, leverage the data argument in the each individual geom (rather that defining this globally) (example below)

R-Ladies Freiburg (English) - Level up your ggplot: Adding labels, arrows and other annotations - YouTube

geom_curve

aes(x,y,xend,yend)

arrow=arrow(length=unit(x,"inch)),

size

colr

curvature(0 is straight line, positive is right hand curve, negative is left hand curve)

ggforce package has advanced annotation options

geom_mark_circle

geom_mark_rect

geom_mark_hull

geom_mark_elipse`

aes(label,filter,description)

expand

label.lineheight

label.fontsize

show.legend

ggforce() package

with_blur() can blur the geoms (ten need to seperately map the geoms that you do want to show

need to warp the geom_jitter comand in with_blur

with_blur(

geom_jitter(),

sigma = unit(#,"mm") #blur impact

facet_zoom()# takes a larger dataset and then adds in a zoomed up graph

facet_zoom(axis =argument==filterar_gument)

facet_zoom(x=country=="spain")

facet_zoom(y=length <20)

Tidy models

data load

rsamples

assign to training df: split

based on proportions and strata

assign to training df: assign trianing

assign cross validation sets (if ncessary)

assign to testingdf: assign testing

initial_split()

training()

testing()

recipes

prepare the data with prepeprosing

assign vairables

other transformation steps

recipte()

update_role()

parsnip

specify and fit the model

model (decision_tree()

set_engine()

set_mode()

workflow

workflow()

add_receipe()

add_model()

Tidy evaluation

Programming with dplyr • dplyr (tidyverse.org)

Argument type: tidy-select — dplyr_tidy_select • dplyr (tidyverse.org)

Tidy evaluation is not all-or-nothing, it encompasses a wide range of features and techniques. Here are a few techniques that are easy to pick up in your workflow:

Passing expressions through {{ and ....

Passing column names to .data[[ and one_of().

All these techniques make it possible to reuse existing comp

When creating forumlas how to referece to names?

STart with fixed names (only if you are sure it wont change) and try wrapping that around a test to ensur eit exists

duoble currly braces {{}}

When you want to reference a data variable from an env. variable in function, you pass the dataframe to the function then wrap the data var with {{var}} in order pull it from the env frame (instead of saying data$x)

where() for search paramters

.data[[var]]

if you env variable is a character frame that you must use .data[[var]]

all_of or any_of for character vectors for search praramters

<!-- -->

compute_bmi <- function(data) {
  if (!all(c("mass", "height") %in% names(data))) {
    stop("`data` must contain `mass` and `height` columns")
  }

  data %>% transmute(bmi = mass / height^2)
}

how to use as nmaes

"mean_{{var}}" := mean({{var}})

Open questions

when do you sue data and when do use .data (okay answer apparently when you use … you start other variables with "." eg. .data to avoid conflictino (20 Dot prefix | Tidyverse design guide

)

f you want the user to provide a set of data-variables that are then transformed, use across():

my_summarise <- function(data, summary_vars) {
  data %>%
    summarise(across({{ summary_vars }}, ~ mean(., na.rm = TRUE)))
}
starwars %>% 
  group_by(species) %>% 
  my_summarise(c(mass, height))
#> # A tibble: 38 × 3
#>   species   mass height
#>   <chr>    <dbl>  <dbl>
#> 1 Aleena      15     79
#> 2 Besalisk   102    198
#> 3 Cerean      82    198
#> 4 Chagrian   NaN    196
#> # … with 34 more rows

You can use this same idea for multiple sets of input data-variables:

my_summarise <- function(data, group_var, summarise_var) {
  data %>%
    group_by(across({{ group_var }})) %>% 
    summarise(across({{ summarise_var }}, mean))
}

Use the .names argument to across() to control the names of the output.

my_summarise <- function(data, group_var, summarise_var) {
  data %>%
    group_by(across({{ group_var }})) %>% 
    summarise(across({{ summarise_var }}, mean, .names = "mean_{.col}"))
}

Action versb to know how to use

Argument type: tidy-select — dplyr_tidy_select • dplyr (tidyverse.org)

everything(): Matches all variables.

last_col(): Select last variable, possibly with an offset.

These helpers select variables by matching patterns in their names:

starts_with(): Starts with a prefix.

ends_with(): Ends with a suffix.

contains(): Contains a literal string.

matches(): Matches a regular expression.

num_range(): Matches a numerical range like x01, x02, x03.

These helpers select variables from a character vector:

all_of(): Matches variable names in a character vector. All names must be present, otherwise an out-of-bounds error is thrown.

any_of(): Same as all_of(), except that no error is thrown for names that don't exist.

This helper selects variables with a function:

where(): Applies a function to all variables and selects those for which the function returns TRUE.

arrange(), count(), filter(), group_by(), mutate(), and summarise() use data masking so that you can use data variables as if they were variables in the environment (i.e. you write my_variable not df$myvariable).

across(), relocate(), rename(), select(), and pull()

rowwise()

colwise()

Tricks

mean in summarize will give you the portion of that variable per the group

purr

resources

9 Basic map functions | Functional Programming (stanford.edu)

Map and Nested Lists | R-bloggers

pattern

take one element .x<-list[[1]]

do the formula based on that element

set_names() without argument sets the names equal to the values

map returns list, control map outcomes with map alternatives eg. map_df, map_dbl

if function has more than one argument then define a function upfront in global environment and pass the second y argument as explicit command in map [follow up how to do this in anonymous way)

map(1:5,custom_function,y=2)

pmap for more than one vector

can also pass through functions as objects not just data

funs<- list(mean,median,sd)

map(funs,~map_dbl(mtcars,.x))

start on the inside and then work your way to the outside

walk similiar ot map but is design for function that you want to run soley for hte side effects

so walk will always return the origional vector eg. walk(.x,.f)=> .x whereas map will return map(.x,.f)=> .f(.x.

So why use walk? when you want the formula side effectt (eg saving a picture)

accumulate

applies same function again and again and again

applies function to first argument then takes that result and applies that outcome to second argument

eg. accumulate(letters,paste), will produce a prymid of values of all the letters

so accumlate will show all the interim values

reduce will only show the final value

this is recursive

However, if you want pair wise actions 1*1, 2*2, etc then you need map2

tidytext

unnest_tokens basically takes a string and breaks t into characters, words, or others ngrams,

From there use typical dplybs to graph, popular geoms are geom_text to plot the words aagainst their proportion.

typiecal tokenize methodologies use ICI (international components of unicode) which defines word boundaries )

Chapter 2 Tokenization | Supervised Machine Learning for Text Analysis in R (smltar.com)

packages

tidytext

tokenize

stopwords

SnowballC for stemming

hunspell also for stemming / spell check

types of toekn

characters

words,

sentences

lines,

paragraphs

ngrams

you can use tidytext package or tokenize package but here is your pattern

grab text

add a dimenion factor (eg. chapter, author, book)

nest the data by the dimension

if data is nested use mutate(map()) pattern to perform transformations

if doing setences or paragraphs you may need to paste() the text and add paragraph breaks "\n" (paragraphs) or space breaks " " (sentences)

then use either tidytext(returns tibble) or tokenizer (returns lists) to do unnesting work

unnest data

anti_join(stop_words)

regex considerations

[:alpha:] brings in non US lettesr where as [a-zA-Z] only brings in US letters

? is optional (will match or not)

^ starts with

$ends with

| or

stop words

stopwords package with snowball, iso and other packages

stemming

can stem words tree, tree's into single word

however also has impact of creating new words to stem by

SnowballC package offers wordStem

tokenizer::tokenize_word_stems

hunspell:hunspell_stem

more resources

Chapter 2 Tokenization | Supervised Machine Learning for Text Analysis in R (smltar.com)

tidytext

tidy evaluation

resources

2 Why and how | Tidy evaluation (tidyverse.org)

Programming with dplyr • dplyr (tidyverse.org)

Implementing tidyselect interfaces • tidyselect (r-lib.org)

Technical description of tidyselect • tidyselect (r-lib.org)

13 Tidy evaluation basics | Functional Programming (stanford.edu)

principles

data masking is when you delay the evaluation of a code so that the code can find relevant columns for computation. th

This is why you can can do::

{r}

#this will work
starwars %>% filter(
  height < 200,
  gender == "male"
)

#but really the program needs this, reference to dataframe and column
starwars[starwars$height < 200 & starwars$gender == "male", ]


technical term for delaying code is quoting

this delaying of code evalution can help you when using code but also make things more complicated when writing code

vectoring can occur when either input has 1 or same length of input object to ensure all columns have same length

some functions can repeat values if recycling completes the length

howevre other,like tidyverse family don't

!! takes a variable defined outside of function and allows you to use it within a function x<-1 function(x) !!x+1, qq_show() allows you to see what is happening

!! works for assiging a variable not a column name or only variable

!! is simliar to := but := is only for left hand side eg setting a name

when working with lists you need !!! to pull out each element and pass it through otherwise !! just pass through the list as is also need to use enquos() vs. enquo

sym) is how you quote for column names and !! is how you unquote, however sym() only works for character strings so "mpg" vs. mpg

enquo is used for non character strings and then !! is how you unquote them

rlang::qq_show will help show what how !! is being evaluated

pattersn

enquo() and !!

:= and !!

enquos() and !!!