inst/extdata/R-code/10-Figures.R
In msperlin/afedR: Data and Functions for Book "Analyzing Financial and Economical Data with R"
#' # Creating and Saving Figures with `ggplot2` {#
#'
## ---- include=FALSE-------------------------------------------------------------------------------------------------
source('Scripts/preamble_chapters.R')
my_fig_height <- 6
my_fig_width <- 7
#'
#' It is a well-known fact that communication is o
#'
#' The easiness of which you can create and presen
#'
#' Visual attractiveness
#' : Always facilitate the analysis of your audien
#'
#' A figure must be self-contained
#' : Any technical information such as origin and
#'
#' The graphical analysis should help your analysi
#' : Just because you can make a plot, doesn't mea
#'
#' Check your references
#' : Science and data analysis evolves as building
#'
#' Said that, let's go back to our R code.
#'
#'
#' ## The `ggplot2` Package
#'
#' R has built-in functions for creating figures,
#'
#' This deficiency was remedied by users. In 2005,
#'
#' In this book, we will not go deep into `ggplot2
#'
#' For most examples given here, we will work with
#'
#' First, let's load the data and check its conten
#'
## -------------------------------------------------------------------------------------------------------------------
library(tidyverse)
# set file and load data
my_f <- afedR::get_data_file('SP500-Stocks-WithRet.rds')
df_sp500 <- read_rds(my_f )
# print first 5 rows
glimpse(df_sp500)
#'
#' It is a fairly common table used in previous ch
#'
#'
#' ## Using Graphics Windows
#'
#' Before studying the use of `ggplot2`, we need t
#'
#' A more intelligent approach to managing figures
#'
## ---- eval=FALSE----------------------------------------------------------------------------------------------------
## x11()
## plot(1:10)
#'
#' The visual result in RStudio should be similar
#'
#'
#' Each call to `x11()` will create a new empty wi
#'
#' After creating so many windows, it is best to c
#'
#'
#' ## Creating Figures with Function `qplot`
#'
#' Package `ggplot2` has an introductory function,
#'
#' To build a time series plot with the prices of
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------------------
library(ggplot2)
# filter stock data
temp_df <- df_sp500 %>%
filter(ticker == 'MMM')
# plot its prices
qplot(data = temp_df,
x = ref.date,
y = price.adjusted,
geom = 'line')
#'
#' In the previous example, the name of the axis c
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------------------
qplot(data = temp_df,
x = ref.date,
y = price.adjusted,
geom = 'line',
xlab = 'Dates',
ylab = 'Adjusted closing prices',
main = 'Prices of MMM')
#'
#' Much better! Notice how the horizontal axis of
#'
#'
#' ## Creating Figures with Function `ggplot` {#gg
#'
#' Using function `qplot` is recommended when you
#'
#' Before presenting examples using `ggplot`, let'
#'
#' The distinction between the steps of creating a
#'
#' Look at the syntax of the following example tha
#'
## ---- eval=TRUE, tidy=FALSE, fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------
p <- ggplot(data = temp_df,
mapping = aes(x = ref.date,
y = price.adjusted))
p <- p + geom_line()
p <- p + labs(x = 'Dates',
y = 'Adjusted closing prices',
title = 'Prices of MMM')
print(p)
#'
#' In using `ggplot`, it is always necessary to pr
#'
#' After defining the input information in argumen
#'
#'
#' Once the data and axis are defined, we save it
#'
## ----eval=TRUE------------------------------------------------------------------------------------------------------
library(ggplot2)
library(stringr)
# get names of functions in ggplot2
fcts <- ls('package:ggplot2')
# select those that starts with geom_
idx <- str_sub(fcts, 1, 5) == 'geom_'
fcts <- fcts[idx]
# print result
print(fcts)
#'
#' As you can see, the `ggplot2` package offers a
#'
#' Going back to our example, the third line of th
#'
#' Using the **pipeline operator** is also possibl
#'
## -------------------------------------------------------------------------------------------------------------------
p <- temp_df %>%
ggplot(aes(x = ref.date, y = price.adjusted)) +
geom_line() +
labs(x = 'Dates',
y = 'Adjusted Closing Prices',
title = 'Prices of MMM')
#'
#' Do notice that we used symbol `+` and **not** `
#'
#' One of the great advantages of using `ggplot` i
#'
## ----eval=TRUE------------------------------------------------------------------------------------------------------
# fix seed
set.seed(10)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
# create temporary df
temp_df <- df_sp500 %>%
filter(ticker %in% tickers)
#'
#' In this code, we use operator `%in%` to find ou
#'
## ----fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------------------------------
p <- temp_df %>%
ggplot(aes(x = ref.date,
y = price.adjusted,
color = ticker)) +
geom_line() +
labs(x = 'Dates',
y = 'Adjusted closing prices',
title = 'Prices of four random stocks',
subtitle = paste0('Date from ', min(temp_df$ref.date),
' to ', max(temp_df$ref.date) ),
caption = 'Data from Yahoo Finance')
print(p)
#'
#' A difference from the previous examples is that
#'
#'
## ----fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------------------------------
p <- p +
scale_color_brewer(palette = "RdYlBu")
print(p)
#'
#' Notice how easy and quick it was to adjust the
#'
#'
#' ### The US Yield Curve
#'
#' Now, let's use what we learned so far to create
#'
#' In the following code, we will first download t
#'
## ---- include=FALSE-------------------------------------------------------------------------------------------------
my_api_key <- 'Esv7Ac7zuZzJSCGxynyF'
#'
## ---- message=FALSE-------------------------------------------------------------------------------------------------
library(GetQuandlData)
library(tidyverse)
# set symbol and dates
my_symbol <- 'USTREASURY/YIELD'
first_date <- as.Date('2010-01-01')
last_date <- Sys.Date()
# get data!
df_yc <- get_Quandl_series(id_in = my_symbol,
first_date = first_date,
last_date = last_date,
api_key = my_api_key)
print(head(df_yc))
#'
#' The result is a dataframe in the **wide format*
#'
## -------------------------------------------------------------------------------------------------------------------
# change to long format and convert to factor
df_yc_long <- df_yc %>%
tidyr::pivot_longer(cols = !ref_date,
names_to = 'maturity',
values_to = 'rate'
) %>%
mutate(rate = as.numeric(rate))
# keep only longer term yields (names with YR)
idx <- str_detect(df_yc_long$maturity, 'YR')
df_yc_long <- df_yc_long[idx, ]
# change name to year number with regex
# obs: regex ([0-9]+) extracts all numbers within a string
out <- str_extract_all(string = df_yc_long$maturity,
pattern = '([0-9]+)')
df_yc_long$maturity <- as.numeric(out)
# glimpse result
glimpse(df_yc_long)
#'
#' We have a dataframe in the long format with thr
#'
## -------------------------------------------------------------------------------------------------------------------
# keep only last date of each
last_date <- max(df_yc_long$ref_date)
df_yc_lastdate <- df_yc_long[df_yc_long$ref_date == last_date, ]
# plot it!
p <- ggplot(df_yc_lastdate, aes(x=maturity, y=rate)) +
geom_point(size = 2.5) + geom_line(size=1) +
labs(x = 'Maturity (years)',
y='Yield Rate (%)',
title = paste0('US Yield Curve for ',last_date),
caption = paste0('Data from Quandl table ', my_symbol, '\n',
'Access at ', Sys.time()))
print(p)
#'
#' As expected, the current yield curve is upward
#'
## -------------------------------------------------------------------------------------------------------------------
# set number of periods
n_periods <- 5
my_year <- 2020
# filter for year 2019
df_yc_my_year <- df_yc_long %>%
filter(lubridate::year(ref_date) == my_year )
# get unique dates in data
unique_dates <- unique(df_yc_my_year$ref_date)
# set sequence of observations
my_seq <- floor(seq(1, length(unique_dates),
length.out = n_periods))
# get actual dates from sequence
my_dates <- unique_dates[my_seq]
# find rows for dates in df
idx <- df_yc_my_year$ref_date %in% my_dates
df_yc_periods <- df_yc_my_year[idx, ]
# plot it!
p <- ggplot(df_yc_periods, aes(x=maturity,
y=rate,
color= factor(ref_date))) +
geom_point(size = 2.5) +
geom_line(size = 1) +
labs(x = 'Maturity (years)',
y='Yield Rate (%)',
title = paste0('US Yield Curve for ', my_year),
color = 'Dates',
caption = paste0('Data from Quandl table ',
my_symbol, '\n',
'Access at ', Sys.time()))
print(p)
#'
#' The yield curve is not static and will change o
#'
#'
#' ## Using Themes
#'
#' One way of customizing graphics in `ggplot2` is
#'
#'
#' Package `ggplot` has a pre-packaged collection
#'
## ----eval=TRUE------------------------------------------------------------------------------------------------------
library(ggplot2)
library(stringr)
# get all functions
fcts <- ls('package:ggplot2')
# find out those that start with theme_
idx <- str_sub(fcts, 1, 6) == 'theme_'
fcts <- fcts[idx]
# print result
print(fcts)
#'
#' Let's try it with the theme from function `them
#'
## ----fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------------------------------
p <- temp_df %>%
ggplot(aes(x = ref.date, y = price.adjusted, color=ticker)) +
geom_line() +
labs(x = 'Dates',
y = 'Adjusted closing prices',
title = 'Prices of four random stocks',
caption = 'Data from Yahoo Finance') +
theme_bw()
print(p)
#'
#' As you can see, the new theme was a white backg
#'
#' Let's now use package `gridExtra` to create a g
#'
## ---- fig.height=9 , fig.width = 9----------------------------------------------------------------------------------
require(gridExtra)
p1 <- p +
theme_bw() +
labs(title = 'Theme BW')
p2 <- p +
theme_dark() +
labs(title = 'Theme Dark')
p3 <- p +
theme_grey() +
labs(title = 'Theme Grey')
p4 <- p +
theme_light() +
labs(title = 'Theme Light')
p5 <- p +
theme_classic() +
labs(title = 'Theme Classic')
p6 <- p +
theme_minimal() +
labs(title = 'Theme Minimal')
grid.arrange(p1, p2, p3,
p4, p5, p6,
ncol=2, nrow = 3)
#'
#'
#' You can try other themes on your computer and s
#'
#' In the previous example, notice how the structu
#'
#' Digging deeper, the selection of the colors fol
#'
## ---- eval=TRUE, tidy=FALSE, fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------
p <- p +
theme_bw() +
scale_colour_grey(start = 0.0, end = 0.6)
print(p)
#'
#' The lines of the plot are now in grey. The inpu
#'
#'
#' ## Creating Panels with `facet_wrap`
#'
#' Another possibility of creating graphics for di
#'
#' Facets are possible with function `facet_wrap`,
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
library(dplyr)
# fix seed
set.seed(10)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ref.date, y = price.adjusted)) +
geom_line() +
labs(x = 'Date',
y = 'Adjusted closing prices',
title = 'Prices of four random stocks',
caption = 'Data from Yahoo Finance') +
facet_wrap(facets = ~ticker) +
theme_bw()
print(p)
#'
#' Using panels is recommended when the data of th
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
# fix seed
set.seed(5)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ref.date, y = ret)) +
geom_line(size=1) +
labs(x = 'Date',
y = 'Returns',
title = 'Daily returns of four random stocks',
caption = 'Data from Yahoo Finance') +
facet_wrap(facets = ~ticker) +
theme_bw()
print(p)
#'
#' Notice how the vertical axis of the panels is f
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------------------
p <- p +
facet_wrap(facets = ~ticker, scales = 'free_y')
print(p)
#'
#' Here, both axis, x and y, have their own scale
#'
#'
#' ## Using the Pipeline
#'
#' We previously saw that `ggplot2` is a friend of
#'
## ---- eval=TRUE, tidy=FALSE,fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------
library(tidyverse)
library(ggplot2)
# calculated mean and sd of returns, plot result
my_f <- afedR::get_data_file(
'SP500_Stocks_long_by_year.rds'
)
df_sp500 <- read_rds(my_f)
p <- df_sp500 %>%
na.omit() %>%
group_by(ticker) %>%
summarise(mean_ret = mean(ret.adjusted.prices),
std_ret = sd(ret.adjusted.prices)) %>%
ggplot(aes(x = std_ret, y = mean_ret)) +
geom_point() +
labs(x = 'Standard Deviation of Yearly returns',
y = 'Average Yearly Returns',
title = 'Expected Return and Risk for SP500 Stocks',
subtitle = paste0('Annual price data from 2010 to 2019, ',
length(unique(df_sp500$ticker)),
' stocks included'),
caption = 'Data imported from Yahoo Finance') +
scale_y_continuous(labels = scales::percent) +
scale_x_continuous(labels = scales::percent) +
theme_bw()
print(p)
#'
#' The previous code is self-contained, easy to re
#'
#' This is another example of an elegant code prod
#'
## Be careful when mixing operators in the `tidyverse`. Whenever passing tables between `dplyr` commands, use the operator `%>%`. In the case of sequencing layers of a `ggplot` chart, always use the sum symbol (`+`).
#'
#'
#' ## Creating Statistical Graphics
#'
#' Package `ggplot` has several options for creati
#'
#'
#' ### Creating Histograms
#'
#' A histogram shows the empirical distribution of
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------------------
# set file and load data
my_f <- afedR::get_data_file('SP500-Stocks-WithRet.rds')
df_sp500 <- read_rds(my_f )
# remove outliers
my_prob_outliers <- 0.01
df_sp500$ret <- afedR::afedR_replace_outliers(
df_sp500$ret,
my_prob = my_prob_outliers
)
# plot the data
p <- ggplot(data = df_sp500, aes(x = ret)) +
geom_histogram(bins = 100) +
labs(y = 'Frequency',
x = 'Returns',
title = paste0('Distribution of returns for ',
'all stocks in the SP500 index'),
subtitle = paste0('Data from 2010 to 2019\n',
'Distribution based quantiles at the ',
scales::percent(my_prob_outliers),
' were removed'),
caption = 'Data from Yahoo Finance'
) +
scale_x_continuous(labels = scales::percent) +
theme_bw()
print(p)
#'
#' Here, we only need to define the _x_ value, wit
#'
#' We can also use groups and facets as we did for
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
# fix seed
set.seed(30)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ret)) +
geom_histogram(bins = 50) +
labs(y = 'Frequency',
x = 'Returns',
title = 'Distribution of returns for four random stocks',
subtitle = paste0('Data from 2010 to 2019\n',
'Quantiles at the ',
scales::percent(my_prob_outliers),
' of the distribution were removed'),
caption = 'Data from Yahoo Finance'
) +
scale_x_continuous(labels = scales::percent) +
theme_bw() +
facet_wrap(facets = ~ticker)
print(p)
#'
#' A histogram with the empirical densities of the
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ret)) +
geom_density() +
facet_wrap(facets = ~ticker) +
labs(y = 'Interpolated Frequency',
x = 'Returns',
title = paste0('Interpolated distribution of returns',
'for all stocks in the SP500 index'),
subtitle = paste0('Data from 2010 to 2019\n',
'Quantiles at the ',
scales::percent(my_prob_outliers),
' of both sides of the distribution',
'were removed'),
caption = 'Data from Yahoo Finance'
) +
scale_x_continuous(labels = scales::percent) +
theme_bw()
print(p)
#'
#' The previous figure allows a clear visual compa
#'
#'
#' ### Creating _boxplot_ Figures
#'
#' Figures of type _boxplot_, or box and whisker d
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
# fix seed
set.seed(30)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ticker, y = price.adjusted)) +
geom_boxplot() +
labs(y = 'Adjusted Price',
x = 'Ticker',
title = paste0('Distribution of daily prices',
' of four random stocks'),
caption = paste0('Data imported from Yahoo Finance',
' (2010 - 2019)')
) +
theme_bw()
print(p)
#'
#' We define a box plot by setting the `x` and `y`
#'
#' Another interesting application of boxplot figu
#'
## -------------------------------------------------------------------------------------------------------------------
p <- ggplot(df_yc_long, aes(x = factor(maturity), y = rate)) +
geom_boxplot() +
labs(title = 'Distribution of Yield Rates over Maturities',
x = 'Maturity (years)',
y = 'Yield Rate (% per year)',
caption = paste0('Data from Quandl \n',
'Created at ', Sys.time()) ) +
theme_bw()
print(p)
#'
#' The statistical plot shows the upward pattern f
#'
#'
#' ### Creating _QQ_ Plots
#'
#' QQ plots show a comparison between the empirica
#'
#' Let's try an example with some simulated data.
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
# fix seed
set.seed(40)
# set options
N <- 10000
my_mean <- 10
my_sd <- 2
# create tibble
temp_df <- tibble(y=rnorm(n = N,
mean = my_mean,
sd = my_sd))
# plot QQ
p <- ggplot(data = temp_df, aes(sample = y)) +
geom_qq(distribution = qnorm,
dparams = c(mean=my_mean, sd=my_sd)) +
labs(title = 'QQ plot of simulated Normal Distribution')
print(p)
#'
#' In the previous code, we simulate random normal
#'
#' Now, let's try it for our table of stock's retu
#'
## ----eval=TRUE,fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE-----------------------------------------
# fix seed
set.seed(10)
# select 4 stock randomly and filter
tickers <- sample(unique(df_sp500$ticker), 4)
temp_df <- df_sp500 %>%
filter(ticker %in% tickers)
# set function for normalization
norm_vec <- function(y){
# Normalizes a vector by subtracting mean and dividing
# by the standard deviation
#
# Args:
# y - numerical vector
#
# Returns:
# A normalized vector
y.norm <- (y-mean(y, na.rm = TRUE))/sd(y, na.rm = TRUE)
return(y.norm)
}
# apply function
my_l <- tapply(X = temp_df$ret,
INDEX = factor(temp_df$ticker),
FUN = norm_vec)
# reorder list (tapply sorts alphabetically)
my_l <- my_l[as.character(unique(temp_df$ticker))]
# save new column norm.ret
temp_df$norm_ret <- unlist(my_l)
# plot it!
p <- ggplot(data = temp_df, aes(sample = norm_ret)) +
geom_qq() +
facet_wrap(~ticker) +
labs(title = 'QQ plot for normalized returns',
subtitle = 'Daily returns from 2010 to 2019',
x = 'Theoretical value (from Normal)',
y = 'True/observed value',
caption = 'Data from Yahoo Finance') +
theme_bw()
print(p)
#'
#'
#' As you can see, the result is not visually simi
#'
#'
#' ## Saving Graphics to a File
#'
#' To save pictures created with `ggplot`, use fun
#'
#' Consider the following example, where we create
#'
## ----eval=TRUE, tidy=FALSE,fig.height=my_fig_height, fig.width=my_fig_width-----------------------------------------
library(tidyverse)
# fix seed
set.seed(40)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ref.date,
y = price.adjusted,
color = ticker)) +
geom_line() +
labs(x = 'Date',
y = 'Adjusted closing prices',
title = 'Prices of four random stocks',
caption = 'Data from Yahoo Finance')
# save file
my_fig_file <- 'fig_ggplot/MyPrices.png'
ggsave(filename = my_fig_file,
plot=p,
dpi = 600)
#'
#' You can verify the creation of the file with fu
#'
## -------------------------------------------------------------------------------------------------------------------
print(list.files('fig_ggplot'))
#'
#' As expected, the file is available in folder `f
#'
#'
#' ## Exercises
#'
## ---- echo=FALSE, results='asis'------------------------------------------------------------------------------------
f_in <- list.files('../02-EOCE-Rmd/Chapter10-Figures/',
full.names = TRUE)
compile_eoc_exercises(f_in, type_doc = my_engine)
#'
#'
msperlin/afedR documentation built on Sept. 11, 2022, 9:49 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
#' # Creating and Saving Figures with `ggplot2` {#
#'
## ---- include=FALSE-------------------------------------------------------------------------------------------------
source('Scripts/preamble_chapters.R')
my_fig_height <- 6
my_fig_width <- 7
#'
#' It is a well-known fact that communication is o
#'
#' The easiness of which you can create and presen
#'
#' Visual attractiveness
#' : Always facilitate the analysis of your audien
#'
#' A figure must be self-contained
#' : Any technical information such as origin and
#'
#' The graphical analysis should help your analysi
#' : Just because you can make a plot, doesn't mea
#'
#' Check your references
#' : Science and data analysis evolves as building
#'
#' Said that, let's go back to our R code.
#'
#'
#' ## The `ggplot2` Package
#'
#' R has built-in functions for creating figures,
#'
#' This deficiency was remedied by users. In 2005,
#'
#' In this book, we will not go deep into `ggplot2
#'
#' For most examples given here, we will work with
#'
#' First, let's load the data and check its conten
#'
## -------------------------------------------------------------------------------------------------------------------
library(tidyverse)
# set file and load data
my_f <- afedR::get_data_file('SP500-Stocks-WithRet.rds')
df_sp500 <- read_rds(my_f )
# print first 5 rows
glimpse(df_sp500)
#'
#' It is a fairly common table used in previous ch
#'
#'
#' ## Using Graphics Windows
#'
#' Before studying the use of `ggplot2`, we need t
#'
#' A more intelligent approach to managing figures
#'
## ---- eval=FALSE----------------------------------------------------------------------------------------------------
## x11()
## plot(1:10)
#'
#' The visual result in RStudio should be similar
#'
#'
#' Each call to `x11()` will create a new empty wi
#'
#' After creating so many windows, it is best to c
#'
#'
#' ## Creating Figures with Function `qplot`
#'
#' Package `ggplot2` has an introductory function,
#'
#' To build a time series plot with the prices of
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------------------
library(ggplot2)
# filter stock data
temp_df <- df_sp500 %>%
filter(ticker == 'MMM')
# plot its prices
qplot(data = temp_df,
x = ref.date,
y = price.adjusted,
geom = 'line')
#'
#' In the previous example, the name of the axis c
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------------------
qplot(data = temp_df,
x = ref.date,
y = price.adjusted,
geom = 'line',
xlab = 'Dates',
ylab = 'Adjusted closing prices',
main = 'Prices of MMM')
#'
#' Much better! Notice how the horizontal axis of
#'
#'
#' ## Creating Figures with Function `ggplot` {#gg
#'
#' Using function `qplot` is recommended when you
#'
#' Before presenting examples using `ggplot`, let'
#'
#' The distinction between the steps of creating a
#'
#' Look at the syntax of the following example tha
#'
## ---- eval=TRUE, tidy=FALSE, fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------
p <- ggplot(data = temp_df,
mapping = aes(x = ref.date,
y = price.adjusted))
p <- p + geom_line()
p <- p + labs(x = 'Dates',
y = 'Adjusted closing prices',
title = 'Prices of MMM')
print(p)
#'
#' In using `ggplot`, it is always necessary to pr
#'
#' After defining the input information in argumen
#'
#'
#' Once the data and axis are defined, we save it
#'
## ----eval=TRUE------------------------------------------------------------------------------------------------------
library(ggplot2)
library(stringr)
# get names of functions in ggplot2
fcts <- ls('package:ggplot2')
# select those that starts with geom_
idx <- str_sub(fcts, 1, 5) == 'geom_'
fcts <- fcts[idx]
# print result
print(fcts)
#'
#' As you can see, the `ggplot2` package offers a
#'
#' Going back to our example, the third line of th
#'
#' Using the **pipeline operator** is also possibl
#'
## -------------------------------------------------------------------------------------------------------------------
p <- temp_df %>%
ggplot(aes(x = ref.date, y = price.adjusted)) +
geom_line() +
labs(x = 'Dates',
y = 'Adjusted Closing Prices',
title = 'Prices of MMM')
#'
#' Do notice that we used symbol `+` and **not** `
#'
#' One of the great advantages of using `ggplot` i
#'
## ----eval=TRUE------------------------------------------------------------------------------------------------------
# fix seed
set.seed(10)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
# create temporary df
temp_df <- df_sp500 %>%
filter(ticker %in% tickers)
#'
#' In this code, we use operator `%in%` to find ou
#'
## ----fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------------------------------
p <- temp_df %>%
ggplot(aes(x = ref.date,
y = price.adjusted,
color = ticker)) +
geom_line() +
labs(x = 'Dates',
y = 'Adjusted closing prices',
title = 'Prices of four random stocks',
subtitle = paste0('Date from ', min(temp_df$ref.date),
' to ', max(temp_df$ref.date) ),
caption = 'Data from Yahoo Finance')
print(p)
#'
#' A difference from the previous examples is that
#'
#'
## ----fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------------------------------
p <- p +
scale_color_brewer(palette = "RdYlBu")
print(p)
#'
#' Notice how easy and quick it was to adjust the
#'
#'
#' ### The US Yield Curve
#'
#' Now, let's use what we learned so far to create
#'
#' In the following code, we will first download t
#'
## ---- include=FALSE-------------------------------------------------------------------------------------------------
my_api_key <- 'Esv7Ac7zuZzJSCGxynyF'
#'
## ---- message=FALSE-------------------------------------------------------------------------------------------------
library(GetQuandlData)
library(tidyverse)
# set symbol and dates
my_symbol <- 'USTREASURY/YIELD'
first_date <- as.Date('2010-01-01')
last_date <- Sys.Date()
# get data!
df_yc <- get_Quandl_series(id_in = my_symbol,
first_date = first_date,
last_date = last_date,
api_key = my_api_key)
print(head(df_yc))
#'
#' The result is a dataframe in the **wide format*
#'
## -------------------------------------------------------------------------------------------------------------------
# change to long format and convert to factor
df_yc_long <- df_yc %>%
tidyr::pivot_longer(cols = !ref_date,
names_to = 'maturity',
values_to = 'rate'
) %>%
mutate(rate = as.numeric(rate))
# keep only longer term yields (names with YR)
idx <- str_detect(df_yc_long$maturity, 'YR')
df_yc_long <- df_yc_long[idx, ]
# change name to year number with regex
# obs: regex ([0-9]+) extracts all numbers within a string
out <- str_extract_all(string = df_yc_long$maturity,
pattern = '([0-9]+)')
df_yc_long$maturity <- as.numeric(out)
# glimpse result
glimpse(df_yc_long)
#'
#' We have a dataframe in the long format with thr
#'
## -------------------------------------------------------------------------------------------------------------------
# keep only last date of each
last_date <- max(df_yc_long$ref_date)
df_yc_lastdate <- df_yc_long[df_yc_long$ref_date == last_date, ]
# plot it!
p <- ggplot(df_yc_lastdate, aes(x=maturity, y=rate)) +
geom_point(size = 2.5) + geom_line(size=1) +
labs(x = 'Maturity (years)',
y='Yield Rate (%)',
title = paste0('US Yield Curve for ',last_date),
caption = paste0('Data from Quandl table ', my_symbol, '\n',
'Access at ', Sys.time()))
print(p)
#'
#' As expected, the current yield curve is upward
#'
## -------------------------------------------------------------------------------------------------------------------
# set number of periods
n_periods <- 5
my_year <- 2020
# filter for year 2019
df_yc_my_year <- df_yc_long %>%
filter(lubridate::year(ref_date) == my_year )
# get unique dates in data
unique_dates <- unique(df_yc_my_year$ref_date)
# set sequence of observations
my_seq <- floor(seq(1, length(unique_dates),
length.out = n_periods))
# get actual dates from sequence
my_dates <- unique_dates[my_seq]
# find rows for dates in df
idx <- df_yc_my_year$ref_date %in% my_dates
df_yc_periods <- df_yc_my_year[idx, ]
# plot it!
p <- ggplot(df_yc_periods, aes(x=maturity,
y=rate,
color= factor(ref_date))) +
geom_point(size = 2.5) +
geom_line(size = 1) +
labs(x = 'Maturity (years)',
y='Yield Rate (%)',
title = paste0('US Yield Curve for ', my_year),
color = 'Dates',
caption = paste0('Data from Quandl table ',
my_symbol, '\n',
'Access at ', Sys.time()))
print(p)
#'
#' The yield curve is not static and will change o
#'
#'
#' ## Using Themes
#'
#' One way of customizing graphics in `ggplot2` is
#'
#'
#' Package `ggplot` has a pre-packaged collection
#'
## ----eval=TRUE------------------------------------------------------------------------------------------------------
library(ggplot2)
library(stringr)
# get all functions
fcts <- ls('package:ggplot2')
# find out those that start with theme_
idx <- str_sub(fcts, 1, 6) == 'theme_'
fcts <- fcts[idx]
# print result
print(fcts)
#'
#' Let's try it with the theme from function `them
#'
## ----fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------------------------------
p <- temp_df %>%
ggplot(aes(x = ref.date, y = price.adjusted, color=ticker)) +
geom_line() +
labs(x = 'Dates',
y = 'Adjusted closing prices',
title = 'Prices of four random stocks',
caption = 'Data from Yahoo Finance') +
theme_bw()
print(p)
#'
#' As you can see, the new theme was a white backg
#'
#' Let's now use package `gridExtra` to create a g
#'
## ---- fig.height=9 , fig.width = 9----------------------------------------------------------------------------------
require(gridExtra)
p1 <- p +
theme_bw() +
labs(title = 'Theme BW')
p2 <- p +
theme_dark() +
labs(title = 'Theme Dark')
p3 <- p +
theme_grey() +
labs(title = 'Theme Grey')
p4 <- p +
theme_light() +
labs(title = 'Theme Light')
p5 <- p +
theme_classic() +
labs(title = 'Theme Classic')
p6 <- p +
theme_minimal() +
labs(title = 'Theme Minimal')
grid.arrange(p1, p2, p3,
p4, p5, p6,
ncol=2, nrow = 3)
#'
#'
#' You can try other themes on your computer and s
#'
#' In the previous example, notice how the structu
#'
#' Digging deeper, the selection of the colors fol
#'
## ---- eval=TRUE, tidy=FALSE, fig.height=my_fig_height, fig.width=my_fig_width---------------------------------------
p <- p +
theme_bw() +
scale_colour_grey(start = 0.0, end = 0.6)
print(p)
#'
#' The lines of the plot are now in grey. The inpu
#'
#'
#' ## Creating Panels with `facet_wrap`
#'
#' Another possibility of creating graphics for di
#'
#' Facets are possible with function `facet_wrap`,
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
library(dplyr)
# fix seed
set.seed(10)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ref.date, y = price.adjusted)) +
geom_line() +
labs(x = 'Date',
y = 'Adjusted closing prices',
title = 'Prices of four random stocks',
caption = 'Data from Yahoo Finance') +
facet_wrap(facets = ~ticker) +
theme_bw()
print(p)
#'
#' Using panels is recommended when the data of th
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
# fix seed
set.seed(5)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ref.date, y = ret)) +
geom_line(size=1) +
labs(x = 'Date',
y = 'Returns',
title = 'Daily returns of four random stocks',
caption = 'Data from Yahoo Finance') +
facet_wrap(facets = ~ticker) +
theme_bw()
print(p)
#'
#' Notice how the vertical axis of the panels is f
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------------------
p <- p +
facet_wrap(facets = ~ticker, scales = 'free_y')
print(p)
#'
#' Here, both axis, x and y, have their own scale
#'
#'
#' ## Using the Pipeline
#'
#' We previously saw that `ggplot2` is a friend of
#'
## ---- eval=TRUE, tidy=FALSE,fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------
library(tidyverse)
library(ggplot2)
# calculated mean and sd of returns, plot result
my_f <- afedR::get_data_file(
'SP500_Stocks_long_by_year.rds'
)
df_sp500 <- read_rds(my_f)
p <- df_sp500 %>%
na.omit() %>%
group_by(ticker) %>%
summarise(mean_ret = mean(ret.adjusted.prices),
std_ret = sd(ret.adjusted.prices)) %>%
ggplot(aes(x = std_ret, y = mean_ret)) +
geom_point() +
labs(x = 'Standard Deviation of Yearly returns',
y = 'Average Yearly Returns',
title = 'Expected Return and Risk for SP500 Stocks',
subtitle = paste0('Annual price data from 2010 to 2019, ',
length(unique(df_sp500$ticker)),
' stocks included'),
caption = 'Data imported from Yahoo Finance') +
scale_y_continuous(labels = scales::percent) +
scale_x_continuous(labels = scales::percent) +
theme_bw()
print(p)
#'
#' The previous code is self-contained, easy to re
#'
#' This is another example of an elegant code prod
#'
## Be careful when mixing operators in the `tidyverse`. Whenever passing tables between `dplyr` commands, use the operator `%>%`. In the case of sequencing layers of a `ggplot` chart, always use the sum symbol (`+`).
#'
#'
#' ## Creating Statistical Graphics
#'
#' Package `ggplot` has several options for creati
#'
#'
#' ### Creating Histograms
#'
#' A histogram shows the empirical distribution of
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width----------------------------------------------------
# set file and load data
my_f <- afedR::get_data_file('SP500-Stocks-WithRet.rds')
df_sp500 <- read_rds(my_f )
# remove outliers
my_prob_outliers <- 0.01
df_sp500$ret <- afedR::afedR_replace_outliers(
df_sp500$ret,
my_prob = my_prob_outliers
)
# plot the data
p <- ggplot(data = df_sp500, aes(x = ret)) +
geom_histogram(bins = 100) +
labs(y = 'Frequency',
x = 'Returns',
title = paste0('Distribution of returns for ',
'all stocks in the SP500 index'),
subtitle = paste0('Data from 2010 to 2019\n',
'Distribution based quantiles at the ',
scales::percent(my_prob_outliers),
' were removed'),
caption = 'Data from Yahoo Finance'
) +
scale_x_continuous(labels = scales::percent) +
theme_bw()
print(p)
#'
#' Here, we only need to define the _x_ value, wit
#'
#' We can also use groups and facets as we did for
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
# fix seed
set.seed(30)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ret)) +
geom_histogram(bins = 50) +
labs(y = 'Frequency',
x = 'Returns',
title = 'Distribution of returns for four random stocks',
subtitle = paste0('Data from 2010 to 2019\n',
'Quantiles at the ',
scales::percent(my_prob_outliers),
' of the distribution were removed'),
caption = 'Data from Yahoo Finance'
) +
scale_x_continuous(labels = scales::percent) +
theme_bw() +
facet_wrap(facets = ~ticker)
print(p)
#'
#' A histogram with the empirical densities of the
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ret)) +
geom_density() +
facet_wrap(facets = ~ticker) +
labs(y = 'Interpolated Frequency',
x = 'Returns',
title = paste0('Interpolated distribution of returns',
'for all stocks in the SP500 index'),
subtitle = paste0('Data from 2010 to 2019\n',
'Quantiles at the ',
scales::percent(my_prob_outliers),
' of both sides of the distribution',
'were removed'),
caption = 'Data from Yahoo Finance'
) +
scale_x_continuous(labels = scales::percent) +
theme_bw()
print(p)
#'
#' The previous figure allows a clear visual compa
#'
#'
#' ### Creating _boxplot_ Figures
#'
#' Figures of type _boxplot_, or box and whisker d
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
# fix seed
set.seed(30)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ticker, y = price.adjusted)) +
geom_boxplot() +
labs(y = 'Adjusted Price',
x = 'Ticker',
title = paste0('Distribution of daily prices',
' of four random stocks'),
caption = paste0('Data imported from Yahoo Finance',
' (2010 - 2019)')
) +
theme_bw()
print(p)
#'
#' We define a box plot by setting the `x` and `y`
#'
#' Another interesting application of boxplot figu
#'
## -------------------------------------------------------------------------------------------------------------------
p <- ggplot(df_yc_long, aes(x = factor(maturity), y = rate)) +
geom_boxplot() +
labs(title = 'Distribution of Yield Rates over Maturities',
x = 'Maturity (years)',
y = 'Yield Rate (% per year)',
caption = paste0('Data from Quandl \n',
'Created at ', Sys.time()) ) +
theme_bw()
print(p)
#'
#' The statistical plot shows the upward pattern f
#'
#'
#' ### Creating _QQ_ Plots
#'
#' QQ plots show a comparison between the empirica
#'
#' Let's try an example with some simulated data.
#'
## ----eval=TRUE, fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE----------------------------------------
# fix seed
set.seed(40)
# set options
N <- 10000
my_mean <- 10
my_sd <- 2
# create tibble
temp_df <- tibble(y=rnorm(n = N,
mean = my_mean,
sd = my_sd))
# plot QQ
p <- ggplot(data = temp_df, aes(sample = y)) +
geom_qq(distribution = qnorm,
dparams = c(mean=my_mean, sd=my_sd)) +
labs(title = 'QQ plot of simulated Normal Distribution')
print(p)
#'
#' In the previous code, we simulate random normal
#'
#' Now, let's try it for our table of stock's retu
#'
## ----eval=TRUE,fig.height=my_fig_height, fig.width=my_fig_width, tidy=FALSE-----------------------------------------
# fix seed
set.seed(10)
# select 4 stock randomly and filter
tickers <- sample(unique(df_sp500$ticker), 4)
temp_df <- df_sp500 %>%
filter(ticker %in% tickers)
# set function for normalization
norm_vec <- function(y){
# Normalizes a vector by subtracting mean and dividing
# by the standard deviation
#
# Args:
# y - numerical vector
#
# Returns:
# A normalized vector
y.norm <- (y-mean(y, na.rm = TRUE))/sd(y, na.rm = TRUE)
return(y.norm)
}
# apply function
my_l <- tapply(X = temp_df$ret,
INDEX = factor(temp_df$ticker),
FUN = norm_vec)
# reorder list (tapply sorts alphabetically)
my_l <- my_l[as.character(unique(temp_df$ticker))]
# save new column norm.ret
temp_df$norm_ret <- unlist(my_l)
# plot it!
p <- ggplot(data = temp_df, aes(sample = norm_ret)) +
geom_qq() +
facet_wrap(~ticker) +
labs(title = 'QQ plot for normalized returns',
subtitle = 'Daily returns from 2010 to 2019',
x = 'Theoretical value (from Normal)',
y = 'True/observed value',
caption = 'Data from Yahoo Finance') +
theme_bw()
print(p)
#'
#'
#' As you can see, the result is not visually simi
#'
#'
#' ## Saving Graphics to a File
#'
#' To save pictures created with `ggplot`, use fun
#'
#' Consider the following example, where we create
#'
## ----eval=TRUE, tidy=FALSE,fig.height=my_fig_height, fig.width=my_fig_width-----------------------------------------
library(tidyverse)
# fix seed
set.seed(40)
# select 4 stocks randomly
tickers <- sample(unique(df_sp500$ticker), 4)
p <- df_sp500 %>%
filter(ticker %in% tickers) %>%
ggplot(aes(x = ref.date,
y = price.adjusted,
color = ticker)) +
geom_line() +
labs(x = 'Date',
y = 'Adjusted closing prices',
title = 'Prices of four random stocks',
caption = 'Data from Yahoo Finance')
# save file
my_fig_file <- 'fig_ggplot/MyPrices.png'
ggsave(filename = my_fig_file,
plot=p,
dpi = 600)
#'
#' You can verify the creation of the file with fu
#'
## -------------------------------------------------------------------------------------------------------------------
print(list.files('fig_ggplot'))
#'
#' As expected, the file is available in folder `f
#'
#'
#' ## Exercises
#'
## ---- echo=FALSE, results='asis'------------------------------------------------------------------------------------
f_in <- list.files('../02-EOCE-Rmd/Chapter10-Figures/',
full.names = TRUE)
compile_eoc_exercises(f_in, type_doc = my_engine)
#'
#'
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.