How do I have multiple lines of the same color with GG Plot

Question

I am using ggplot to plot production for over time by gas well.


GAS_PRODUCTION_CURVE <- RawdataTest %>% ggplot(mapping=aes(x=DaysOn, y=GasProd_MCF, color=WellID)) + 
  geom_line(size=0.5) + theme_bw() + 
  scale_color_manual(values = cols) + scale_y_continuous(label=comma) + 
  coord_cartesian(xlim = c(0, max(RawdataTest$DaysOn)), ylim = c(0,max(RawdataTest$GasProd_MCF))) + 
  theme(legend.position="none") + xlab("Days On") +
  ylab("Gas Rate [MCF]")

This gets me the plot that I want (note: this is just a subset of the data). However, I want to have the well data graphed, but color by the variable "RSOperator". In other words, I want all wells with the same RSOperator to be the same color. That way users can distinguish the difference in well performance between wells. Is there a way to adjust my code to accomplish this?

Check this out sape.inf.usi.ch/quick-reference/ggplot2/colour — IceeFrog
– IceeFrog, Commented Nov 21, 2019 at 13:30
If I eliminate color=WellID, how do I keep the well level data graphed but color by RSOperator? — Tim Batten
– Tim Batten, Commented Nov 21, 2019 at 13:38

StupidWolf · Accepted Answer · 2019-11-21 15:21:00Z

I simulate some data that hopefully looks like yours, and you can see how to get the same color for a common RSOperator.

RawdataTest = data.frame(
  DaysOn = rep(1:10,6),
  GasProd_MCF = c(rep(1:10,3),rep(2*(1:10),3))+rnorm(60,3,1),
  WellID = rep(1:3,each=10,times=2),
  RSOperator = rep(letters[1:2],each=30)
)
# create a uniq identifier for observation
RawdataTest <- RawdataTest %>% 
mutate(uniq_id=paste(RSOperator,WellID,sep=""))

# create mapping for uniq id to color, depends on RSOperator
MAPPING <- RawdataTest %>% distinct(RSOperator,uniq_id)
RS_COLS =  brewer.pal(9,"Set1")
RS_COLS = RS_COLS[1:n_distinct(MAPPING$RSOperator)]
names(RS_COLS) = unique(MAPPING$RSOperator)
PLOT_COLS = RS_COLS[MAPPING$RSOperator]
names(PLOT_COLS) = MAPPING$uniq_id

 ggplot(RawdataTest,mapping=aes(x=DaysOn, y=GasProd_MCF,col=uniq_id)) + 
  geom_line(size=0.5) + theme_bw() + 
  scale_color_manual(values = PLOT_COLS)

However you can see it's hard to distinguish a1 from a2 etc.. You might need to consider combining it with linetype, but once you have a lot of lines, it gets crazy:

LINETYPE = rep(1:3,2)
names(LINETYPE) = MAPPING$uniq_id

 ggplot(RawdataTest,mapping=aes(x=DaysOn, y=GasProd_MCF,linetype=uniq_id,col=uniq_id)) + 
  geom_line(size=0.5) + theme_bw() + 
  scale_color_manual(values = PLOT_COLS) +
  scale_linetype_manual(values=LINETYPE)

Grigory Zhadko · Accepted Answer · 2021-04-14 14:10:11Z

From what I learned from StupidWolf's answer, we just need to input a named vector into:

ggplot(RawdataTest,mapping=aes(x=DaysOn, y=GasProd_MCF,col=uniq_id)) + 
  geom_line(size=0.5) + theme_bw() + 
  scale_color_manual(values = NAMED_VECTOR)

Where each element is a color and its name is a value of the column that identifies each line of the plot. The same applies to linetypes.

I took the liberty of creating a function that is heavily inspired by what StupidWolff did in his answer, but which makes more use of the %>% operator for clarity and has some added functionality (as allowing to specify a color for each value). I must admit that the function started being way shorter.

#id: character vector with the id variable
#variable: character vector with the variable we are going to use for styling
#styles: style options (Preferably the same length or longer as unique variable values - Throws a warning)
#named_style: TRUE or FALSE. Specifies whether to create a named vector from values and styles.
#values: values from a variable for which styles are specified. Must be in the same order than each respective style. Other values will be styled according to other_styles.
#other_styles: applies when values specified. Specifies styles for other (non specified) values.


#Plot style mapper - Generates a named vector that can be used for Ggplot styles and colors
ggplot_style_mapper <- function(df, id, variable, styles, values = NULL, other_styles = NULL) {

variablequo <- enquo(variable)
#Style_by_variable
style_by_variable <- if(is.null(values)) {FALSE} else {TRUE}

#Warning
if((n_distinct(df[[variable]]) > length(styles))&style_by_variable == FALSE) {warning("style vector is shorter than unique id-variables")}
styles <- if(style_by_variable == TRUE&length(styles) > length(values)) {styles[1:length(values)]} else (styles)

#Other styles
other_styles <- if(!is.null(other_styles)) {other_styles[!other_styles %in% styles]} else {NULL}
if((length(other_styles) == 0|is.null(other_styles))&style_by_variable == TRUE&(length(values) < length(unique(df[[variable]])))) {warning("Either other_styles necessary but not specified, or other_styles %in% styles")}

#Named_style = TRUE
named_vector <- if(style_by_variable == TRUE) {
 
  mapped <- df %>%
    distinct_at(.vars = c(variable)) %>%
    filter(., .data[[!!variablequo]] %in% values) %>%
    {if(nrow(.) > length(styles)) add_column(., style = c(rep(styles, length.out = nrow(.)))) else
      add_column(., style = styles[1:nrow(.)])} 
  
  dataframe <- df %>%
    distinct_at(., .vars = c(id, variable)) %>%
    select(., all_of(c(id, variable))) %>%
    left_join(., mapped, by = variable)
  
  NAs <- dataframe %>%
    filter(., is.na(style)) %>%
    select(., all_of(c(id, variable))) %>% 
    {if(nrow(.) == 0) . else if (nrow(.) > length(other_styles)) add_column(., style = c(rep(other_styles, length.out = nrow(.)))) else
      add_column(., style = other_styles[1:nrow(.)])} 
  
  dataframe %>%
    filter(., !is.na(style)) %>%
    bind_rows(., NAs) %>%
    pull(., .data[["style"]], name = .data[[id]])
  
  } else {
      
    mapped <- df %>%
      distinct_at(.vars = c(variable)) %>%
      {if(nrow(.) > length(styles)) add_column(., style = c(rep(styles, length.out = nrow(.)))) else
        add_column(., style = styles[1:nrow(.)])} 
    
    dataframe <- df %>%
      distinct_at(., .vars = c(id, variable)) %>%
      select(., all_of(c(id, variable))) %>%
      left_join(., mapped, by = variable) %>%
      pull(., .data[["style"]], name = .data[[id]])}

named_vector }

The function allows us to more easily define the named vector with the styling before creating the graph:

#Create named vectors and graph
#Create named vectors 
named_colors <- ggplot_style_mapper(RawdataTest, id = "uniq_id", variable = "RSOperator",
                                    styles = RColorBrewer::brewer.pal(9,"Set1"))
named_linetype <- ggplot_style_mapper(RawdataTest, id = "uniq_id", variable = "WellID",
                                      styles = c(1,2,3))

#Graph
ggplot(RawdataTest,mapping=aes(x=DaysOn, y=GasProd_MCF,linetype=uniq_id,col=uniq_id)) + 
  geom_line(size=0.5) + theme_bw() + 
  scale_color_manual(values = named_colors) +
  scale_linetype_manual(values= named_linetype)

We could also do it in one step:

#Alternatively, create the graph in one step (shorter but messier)
ggplot(RawdataTest,mapping=aes(x=DaysOn, y=GasProd_MCF,col=uniq_id, linetype=uniq_id)) + 
  geom_line(size=0.5) + theme_bw() + 
  scale_color_manual(values = ggplot_style_mapper(RawdataTest, id = "uniq_id", variable = "RSOperator",
                                                  styles = RColorBrewer::brewer.pal(9,"Set1"))) +
   scale_linetype_manual(values= ggplot_style_mapper(RawdataTest, id = "uniq_id", variable = "WellID",
                                                styles = c(1,2,3)))

The extra functionality I added was that the function allows us to specify the styling of each specific value of the relevant variable. It even allows us to specify styling for just a few values and groups the rest into one styling. Finally, we can specify styling for one or a few values and assign a pallette of stylings to the rest.

#Style for each specific value
named_colors <- ggplot_style_mapper(df = RawdataTest, id = "uniq_id", variable = "RSOperator",
                                    styles = c("blue", "red"), values = c("a", "b"))
named_linetype <- ggplot_style_mapper(df = RawdataTest, id = "uniq_id", variable = "WellID",
                                      styles = c(3,2,1), values = c(1,2,3))

#Style for a few values and other style for the rest
named_colors <- ggplot_style_mapper(df = RawdataTest, id = "uniq_id", variable = "RSOperator",
                                    styles = c("blue"), values = c("a"), other_styles = "black")
named_linetype <- ggplot_style_mapper(df = RawdataTest, id = "uniq_id", variable = "WellID",
                                      styles = c(3), values = c(1,2), other_styles = 6)

#Style for a few values and style pallette for the rest
named_colors <- ggplot_style_mapper(df = RawdataTest, id = "uniq_id", variable = "RSOperator",
                                    styles = c("blue"), values = c("a"), other_styles = c("black"))
named_linetype <- ggplot_style_mapper(df = RawdataTest, id = "uniq_id", variable = "WellID",
                                      styles = c(3), values = c(1), other_styles = c(6,7))

Hope this helps somebody!

pd: data

pacman::p_load(RColorBrewer, tidyverse)

#Create dataframe as StupidWolf
#Create data
RawdataTest = data.frame(
  DaysOn = rep(1:10,6),
  GasProd_MCF = c(rep(1:10,3),rep(2*(1:10),3))+rnorm(60,3,1),
  WellID = rep(1:3,each=10,times=2),
  RSOperator = rep(letters[1:2],each=30)
)
# create a uniq identifier for observation
RawdataTest <- RawdataTest %>% 
  mutate(uniq_id=paste(RSOperator,WellID,sep=""))

Collectives™ on Stack Overflow

How do I have multiple lines of the same color with GG Plot

2 Answers 2

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related