Plotting n columns of a data frame as lines with ggplot in r

Question

I am trying to recreate the Pólya urn model (https://en.wikipedia.org/wiki/Pólya_urn_model) in R, with ggplot. The model basically starts with 1 white and 1 black ball in an 'urn' and randomly chooses one ball and put it back together with a ball of the same color. I do this in R for lets say 10 iterations (so 10 times take out one ball and put it back together with another ball from the same color). And I run this say 5 times. Thus, i get a data frame of 5 columns (=for each run) and 10 rows (=for the iterations).

What I want is to illustrate is this but then this picture has a lot more trials and iterations obviously.

What I have so far is a data frame where each column is the fraction of white balls in the urn per trial/run and I would like to illustrate how the proportions changed for each iteration. And I want to show this separately for each run, so each run I would like to have in a different color.

I have looked through countless similar questions but did not find an answer. I think it's because my data frame has now 5 columns but when i reshape it then I get only a single column of the proportions and next to each I get a code illustrating which column it belonged to - but in this case ggplot only draws one single line in 4 colors.

my data frame looks like this:
          V1         V2         V3        V4 id
1  0.3333333 0.33333333 0.33333333 0.3333333  1
2  0.5000000 0.25000000 0.25000000 0.2500000  2
3  0.4000000 0.20000000 0.20000000 0.4000000  3
4  0.3333333 0.16666667 0.16666667 0.3333333  4
5  0.2857143 0.14285714 0.14285714 0.2857143  5
6  0.2500000 0.12500000 0.12500000 0.3750000  6
7  0.2222222 0.11111111 0.11111111 0.3333333  7
8  0.2000000 0.10000000 0.10000000 0.3000000  8
9  0.1818182 0.09090909 0.09090909 0.2727273  9
10 0.2500000 0.08333333 0.08333333 0.2500000 10

but to make it easier here's some test code:

V1 <- rnorm(10, 0.5, 0.1)
V2 <- rnorm(10, 0.5, 0.1)
V3 <- rnorm(10, 0.5, 0.1)
V4 <- rnorm(10, 0.5, 0.1)
df <- data.frame(V1, V2, V3, V4)

My code for the ggplot is the following :

library(reshape2)
df$id = row.names(df) # add id to each row 
df_long = melt(df, id.vars = "id")  # reshape the data into long format

this first version only depicts the points

ggplot(df_long, aes(x = value, y = id, color = variable)) + 
geom_point()

and this version somehow gets the lines 'messed up' and i cannot figure out why.

ggplot() + geom_line(data = df_long, aes(x = value, y = id, color = variable, group = variable)) + xlab("x axis") +  ylab("y axis")

Any help would be appreciated, I've been really struggling for days with this and couldn't make any significant breakthroughs so far.

EDIT: By 'messed up' I mean that instead of plotting one line per run (which I want to get), the data points seem to lose which trial/run they belong to. So instead of getting one line per run/trial, i get more lines from which some only connect 2-3 points and often connect points from different runs. I hope my explanation is clear enough.

Can you define 'messed up'? I am seeing a graph with one line per value of variable, V1 to V4. — Rui Barradas
– Rui Barradas, Commented May 18, 2018 at 11:57

Ankur Sinha · Accepted Answer · 2018-05-18 13:02:32Z

3

This seems to connect all of them correctly if I understood you correctly. Please check if this is correct.

df$id = 1:nrow(df)
final_data <- melt(df, id='id')
names(final_data) <- c('id', 'func', 'value')

ggplot() + geom_line(data = final_data, aes(x = id, y = value, color = func, group = func), size = 1)

Output:

          V1        V2        V3        V4 id
1  0.4656275 0.4846357 0.4613710 0.5885883  1
2  0.4312952 0.4929042 0.5499502 0.5133333  2
3  0.5890201 0.4652452 0.5598206 0.4789956  3
4  0.7108441 0.4143140 0.5738660 0.4073124  4
5  0.6374072 0.6671785 0.5111608 0.4475132  5
6  0.4797948 0.6191391 0.5423101 0.4472512  6
7  0.5868793 0.5601147 0.4369428 0.5696494  7
8  0.5169970 0.4398982 0.5137524 0.3923140  8
9  0.3960616 0.3552303 0.4174657 0.4449402  9
10 0.5222120 0.5028562 0.5760920 0.4310323 10

answered May 18, 2018 at 13:02

Ankur Sinha

6,6648 gold badges44 silver badges74 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

patL · Accepted Answer · 2018-05-18 11:59:01Z

2

Using your df you can do something like that:

library(tidyverse)

# I use 'gather' instead of 'melt'
df_long = df %>% 
  mutate(id = 1:nrow(.)) %>% 
  gather(id.vars, values, -id) 

df_long %>% 
  ggplot(aes(x = values, y = id, group = id.vars, color = id.vars)) + 
  geom_line(size = 1)

![enter image description here ]

Obs.:

if you set.seed(...) we can replicate your df object.

answered May 18, 2018 at 11:59

patL

2,2971 gold badge20 silver badges39 bronze badges

2 Comments

Roni Over a year ago

thank you! this worked although i had to change it slightly because for some reason the values for the two axes needed to be changed (so: x = id, y = values) but in the end it works so that's what matters.

patL Over a year ago

That's great @Ron

Collectives™ on Stack Overflow

Plotting n columns of a data frame as lines with ggplot in r

2 Answers 2

Comments

Obs.:

2 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Obs.:

2 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related