CI on cumulated values

Question

I have measured the emission of acetone and acetaldehydefrom twosoil_types: AandBthree times across 94days```.

I have calculated and visualized the cumulated emission over time, however I am having issues putting CI 95% Bands (+/-2 SD) or even simple error bars would do.

This is what I've managed to do:

df_cum <- df%>%
  group_by(soil_type, compound, days)%>%
  summarise(mean=mean(emission))%>%
  mutate(cum_emission=cumsum(mean)) 


  plot <- ggplot(df_cum, aes(x = days, y = cum_emission, colour=soil_type)) + 
    geom_line(size = 1)+
    geom_point()+
    scale_colour_manual(values=c("#00AFBB", "brown")) + 
    scale_size_manual(values=c(8.4, 1.7))+
    labs(x = "Time (days)", 
         y = "Cumulated production (umol/g dw soil)", 
         title = "Cumulated production") + 
    labs(shape="", color="") +
    facet_wrap(~compound)+
    theme_bw()

Giving this:

I'm hoping to do be able to put error bars in my plot or do a plot similar to this one with CI:

The data looks like this:

df <- structure(list(days = c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 10, 
10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 
10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 94, 
94, 94, 94, 94, 94, 94, 94, 94, 94, 94, 94, 94, 94, 94, 94, 94, 
94, 94, 94, 94, 94, 94, 94, 94, 94), soil = c(6, 6, 12, 12, 2, 
2, 1, 1, 14, 14, 4, 4, 33, 33, 38, 38, 34, 34, 37, 37, 36, 36, 
13, 13, 32, 32, 5, 5, 3, 3, 35, 35, 6, 6, 12, 12, 2, 2, 1, 1, 
14, 14, 4, 4, 33, 33, 38, 38, 34, 34, 37, 37, 36, 36, 13, 13, 
32, 32, 5, 5, 3, 3, 35, 35, 6, 6, 12, 12, 2, 2, 4, 4, 33, 33, 
38, 38, 34, 34, 37, 37, 36, 36, 13, 13, 5, 5, 3, 3, 35, 35), 
    soil_type = c("B", "B", "A", "A", "B", "B", "B", "B", "A", 
    "A", "B", "B", "B", "B", "B", "B", "A", "A", "B", "B", "A", 
    "A", "A", "A", "B", "B", "A", "A", "B", "B", "B", "B", "B", 
    "B", "A", "A", "B", "B", "B", "B", "A", "A", "B", "B", "B", 
    "B", "B", "B", "A", "A", "B", "B", "A", "A", "A", "A", "B", 
    "B", "A", "A", "B", "B", "B", "B", "B", "B", "A", "A", "B", 
    "B", "B", "B", "B", "B", "B", "B", "A", "A", "B", "B", "A", 
    "A", "A", "A", "A", "A", "B", "B", "B", "B"), compound = c("Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
    "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
    "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde"), emission = c(0.001, 
    0.001, 0.009, 0.004, 0.029, 0.032, 0.066, 0.057, 0.015, 0.015, 
    0, 0.003, 0.015, 0.011, 0.016, 0.005, 0.046, 0.011, 0.004, 
    0.005, 0.015, 0.003, 0.025, 0.012, 0.001, 0.001, 0.004, 0, 
    0.012, 0.002, 0.003, 0.002, 0.006, 0, 0.008, 0.001, 0.061, 
    0.055, 0.076, 0.056, 0.056, 0.074, 0, 0, 0.018, 0.02, 0.015, 
    0.001, 0.064, 0, 0.012, 0.004, 0.009, 0, 0.399, 0.037, 0.002, 
    0.001, 0.116, 0, 0.139, 0, 0.004, 0.001, 0, 0, 0, 0, 0, 0, 
    0, 0, 0, 0, 0, 0, 0.001, 0, 0, 0, 0, 0, 0, 0, 0.001, 0.005, 
    0.001, 0, 0.002, 0)), row.names = c(NA, -90L), class = c("tbl_df", 
"tbl", "data.frame"))

new_df <- structure(list(daysincubated4 = c(0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 4L, 4L, 4L, 4L, 4L, 
4L, 4L, 4L, 4L, 4L, 4L, 4L, 10L, 10L, 10L, 10L, 10L, 10L, 10L, 
10L, 10L, 10L, 10L, 10L, 10L, 10L, 10L, 10L, 10L, 10L, 4L, 4L, 
4L, 4L, 4L, 4L), soil = c(6L, 6L, 12L, 12L, 2L, 2L, 1L, 1L, 14L, 
14L, 4L, 4L, 13L, 13L, 5L, 5L, 3L, 3L, 6L, 6L, 12L, 12L, 2L, 
2L, 1L, 1L, 14L, 14L, 4L, 4L, 6L, 6L, 12L, 12L, 2L, 2L, 1L, 1L, 
14L, 14L, 4L, 4L, 13L, 13L, 5L, 5L, 3L, 3L, 13L, 13L, 5L, 5L, 
3L, 3L), soil_type = c("SOC<10", "SOC<10", "SOC>10", "SOC>10", 
"SOC<10", "SOC<10", "SOC<10", "SOC<10", "SOC>10", "SOC>10", "SOC<10", 
"SOC<10", "SOC>10", "SOC>10", "SOC>10", "SOC>10", "SOC<10", "SOC<10", 
"SOC<10", "SOC<10", "SOC>10", "SOC>10", "SOC<10", "SOC<10", "SOC<10", 
"SOC<10", "SOC>10", "SOC>10", "SOC<10", "SOC<10", "SOC<10", "SOC<10", 
"SOC>10", "SOC>10", "SOC<10", "SOC<10", "SOC<10", "SOC<10", "SOC>10", 
"SOC>10", "SOC<10", "SOC<10", "SOC>10", "SOC>10", "SOC>10", "SOC>10", 
"SOC<10", "SOC<10", "SOC>10", "SOC>10", "SOC>10", "SOC>10", "SOC<10", 
"SOC<10"), compound = c("Acetone", "Acetaldehyde", "Acetone", 
"Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
"Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
"Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
"Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
"Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
"Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
"Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
"Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
"Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", 
"Acetone", "Acetaldehyde", "Acetone", "Acetaldehyde", "Acetone", 
"Acetaldehyde"), emission = c(0.0011, 0.001, 0.0094, 0.0042, 
0.0289, 0.0318, 0.0656, 0.0571, 0.0152, 0.0146, 0, 0.0033, 0.0247, 
0.0117, 0.0038, 3e-04, 0.0124, 0.0016, 8e-04, 1e-04, 0.0188, 
0.0139, 0.0728, 0.0818, 0.0883, 0.0731, 0.0251, 0.0774, 0, 0, 
0.0061, 2e-04, 0.0084, 7e-04, 0.061, 0.0551, 0.0761, 0.0559, 
0.0563, 0.0742, 0, 0, 0.3989, 0.0367, 0.1163, 4e-04, 0.1386, 
4e-04, 0.0461, 0.0495, 0.089, 0.001, 0.1414, 0.001)), row.names = c(NA, 
-54L), class = c("tbl_df", "tbl", "data.frame"))

user_12258459 · Answer

If I understood it correctly, you basically need to estimate the confidence interval for your cumulative line for soil type in different compound, if we look at your data:

ggplot(df,aes(x=days,y=cumsum(emission),col=soil_type)) + 
geom_line() + facet_grid(compound ~ soil) + theme_bw()

What you want to do is estimate the error across lines of the same color in the same row (compound). To do this, we exclude 1 and 14 from your example since it is incomplete. Basically we are trying to convert all your measurements to cumulative, grouped by soil x compound.

We can use stat_summary() to calculate the mean and se:

df %>% filter(!soil %in% c(1,14)) %>% 
group_by(soil,compound) %>% 
mutate(c_emission = cumsum(emission)) %>% 
ggplot(aes(x = days, y = c_emission, col=soil_type,fill=soil_type)) + 
stat_summary(geom="line", fun = mean) + 
stat_summary(geom="ribbon",alpha=0.05,linetype="dotted")+ 
facet_wrap(~compound)

What we are plotting above is the mean at every time point, + 1* standard error. If you want the 95% confidence interval

df %>% filter(!soil %in% c(1,14)) %>% 
group_by(soil,compound) %>% 
mutate(c_emission = cumsum(emission)) %>% 
ggplot(aes(x = days, y = c_emission, col=soil_type,fill=soil_type)) +
stat_summary(geom="line",fun = mean) + 
stat_summary(geom="ribbon",alpha=0.05,linetype="dotted",
fun.data = mean_cl_normal)+ 
facet_wrap(~compound)

The confidence interval is huge because your standard error is large.. Might be better when you have more samples.

user_14215626 · Answer

To plot error bars, have you already tried with

geom_errorbar(aes(ymin = y- 2*sd, xmax = y+ 2*sd), width = .1, position=position_dodge(.9))

with sd being your standard deviation? you can calculate it in the mutate(...) part of the code. I am afraid, though, that you don't have enough replicates to calculate sd/CI. Am I missing something?

CI on cumulated values

2 Answers