In LudvigOlsen/LRO.utilities: Utility functions and addins

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)

LRO.utilities

R package: Utility functions and addins for RStudio.

Helper tools I use in many projects.

By Ludvig R. Olsen,
Cognitive Science, Aarhus University.
Started in Feb. 2017

Contact at:
[e-mail]

Main functions:

• polynomializer
• %ni%
• insertPipe (Addin)
• scaler, scaler_, scaler_fit, scaler_fit_, scaler_transform, scaler_invert
• binarizer, binarizer_
• rename_col
• savage_dickey
• roll_previous

Installation

Development version:

install.packages("devtools")

devtools::install_github("LudvigOlsen/LRO.utilities")

Use

Addins

• Install package.
• Add key command by going to:
  • Tools > Addins > Browse Addins > Keyboard Shortcuts.
  • Find Insert %>% and press its field under Shortcut.
  • Press desired key command.
  • Press Apply.
  • Press Execute.
• Press chosen key command inside R Markdown document.

Examples

# Attach packages
library(LRO.utilities)
library(dplyr) # %>% 
library(knitr) # kable()
library(ggplot2)

---

polynomializer

Exponentiate vectors to make polynomials of degree 2 to degree.

E.g. create columns v^2, v^3, v^4...

# On vector

vect <- c(1,3,5,7,8)

polynomializer(vect, degree = 3) %>% 
  kable()

# On vectors in dataframe

data <- data.frame(vect = vect,
                   bect = vect*3,
                   dect = vect*5)

polynomializer(data, 
               cols = c('bect','dect'), 
               degree = 3) %>% 
  kable()

---

%ni%

"Not in"

c(1,3,5) %ni% c(2,3,6)

---

scaler

Center and/or scale multiple columns of a dataframe.

scaler is designed to work with %>% pipelines.

scaler_fit returns fit_object with information used to transform data.

scaler_transform scales data based on the information in the fit_object.

scaler_invert inverts scaling based on the information in the fit_object.

scaler_ and scaler_fit_ are standard evalution versions.

# Scale and center 'vect' and 'bect' 
# in dataframe from previous example
scaler(data, vect, bect) %>% 
  kable()

# Only scaling 'vect'  - working in pipeline
data %>% 
  scaler(vect, center = F) %>% 
  kable()

# Only center 'bect' and 'dect' 
# selecting with column index range
data %>% 
  scaler(2:3, scale = F) %>% 
  kable()

Fit / Transform / Invert

# Fit scaler
fitted_scaler <- data %>% 
  scaler_fit()

fitted_scaler %>% kable()

# Transform data
scaled_df <- data %>% 
  scaler_transform(fit_object = fitted_scaler)

scaled_df %>% kable()

# Invert scaling
scaled_df %>% 
  scaler_invert(fit_object = fitted_scaler) %>% 
  kable()

---

binarizer

Binarize multiple columns of a dataframe based on a given threshold.

binarizer is designed to work with %>% pipelines.

binarizer_ is a standard evalution version.

scaled_df %>% 
  binarizer(thresh = 0)

---

rename_col

Rename single column in dataframe. This is a bit like plyr::rename just only for 1 column at a time.

rename_col(data, 
           old_name = 'bect', 
           new_name = 'sect') %>% 
  kable()

savage_dickey

Calculate Bayes factor from 2 distributions and plot the the two distributions.

Returns list with ggplot2 object, BF10, and BF01.

prior <- rnorm(1000, mean=0, sd=1)
posterior <- rnorm(1000, mean=2, sd=2)

s_d <- savage_dickey(posterior, prior, Q = 0, plot = TRUE)

s_d$BF10

s_d$BF01

s_d$post_prior_plot +
  theme_bw()

roll_previous

Wrapper for zoo::rollapply for applying a function to rolling windows and getting the result of the previous window. Appends NAs at start only.

# Create dataframe
df <- data.frame('round' = c(1,2,3,4,5,6,7,8,9,10),
                 'score' = c(5,3,4,7,6,5,2,7,8,6))

# For each row we find the mean score of the previous 2 rounds
df$mean_prev_score = roll_previous(df$score, width = 2, FUN = mean)

df %>% kable()

LudvigOlsen/LRO.utilities documentation built on Nov. 13, 2020, 11:26 a.m.