not enough data to fit regression

pfarelo · June 18, 2021, 2:11pm

greetings,

i would like to fit a regression line on grouped data in the graph provided to kind of like see the trend or model with the increase in concentration/dose. Some equation seem to be displaying but no actual line on the graph to show the regression line. I have used the code below and getting error message . I would like to display the equation of the different varieties and the R square thereof Rplot02

library(ggplot2)

library(ggpmisc)
formula1 <- avg_AB ~ Dose
ggplot(data = MeanSE_AB, aes(x = Dose, y = avg_AB)) +
geom_point() +
geom_smooth(method = "lm", se = FALSE, formula = formula1) +
stat_poly_eq(aes(label = paste0("atop(", ..eq.label.., ",", ..rr.label.., ")")),
formula = formula1,
parse = TRUE) +
theme_bw(base_size = 16)

Warning messages:
1: Not enough data to perform fit for group 1; computing mean instead.
2: Not enough data to perform fit for group 2; computing mean instead.
3: Not enough data to perform fit for group 3; computing mean instead.

startz · June 18, 2021, 2:36pm

Perhaps you could show us the data in group 1 of MeanSE_AB?

startz · June 18, 2021, 6:15pm

I did lm() on the first four observations, which worked fine. Check if "group 1" is in fact the first four observations maybe?

pfarelo · June 18, 2021, 6:19pm

yes , the 1st four observations are group 1 and so on..
what do you mean you did lm()?, does that mean this was done to all groups?

startz · June 18, 2021, 6:25pm

foo
# A tibble: 4 x 4
  Variety  Dose avg_AB    se
  <chr>   <dbl>  <dbl> <dbl>
1 Akwa       25   13.2  5.07
2 Akwa       50   30.8  6.98
3 Akwa       75   24.8  4.9 
4 Akwa      100   43.2  5.93```

lm(avg_AB ~ Dose,data = foo)

Call:
lm(formula = avg_AB ~ Dose, data = foo)

Coefficients:
(Intercept) Dose
7.000 0.336

So it looks to me that you DO have enough data, so the bug is not what it appears to be.

pfarelo · June 18, 2021, 6:34pm

what i am actually looking for is the lm () equation of the group represented in the graph and indicating the R2 value as in the attached Example

startz · June 18, 2021, 6:53pm

Your code looks correct to me, which I understand is not very helpful.

FJCC · June 18, 2021, 7:10pm

I don't know how to adjust the positions of the regression equations in the first version so I added a faceted version also.

library(ggplot2)
library(ggpmisc)
MeanSE_AB <- read.csv("~/R/Play/Dummy.csv",sep = " ")
formula1 <- avg_AB ~ Dose
ggplot(data = MeanSE_AB, aes(x = Dose, y = avg_AB,color=Variety)) +
  geom_point() +
  geom_smooth(method = "lm", se = FALSE, formula = y~x) +
  stat_poly_eq(aes(label = paste0("atop(", ..eq.label.., ",", ..rr.label.., ")")),
               formula = y~x,
               parse = TRUE) +
  theme_bw(base_size = 16)


ggplot(data = MeanSE_AB, aes(x = Dose, y = avg_AB)) +
  geom_point() +
  facet_wrap(~Variety)+
  geom_smooth(method = "lm", se = FALSE, formula = y~x) +
  stat_poly_eq(aes(label = paste0("atop(", ..eq.label.., ",", ..rr.label.., ")")),
               formula = y~x,
               parse = TRUE) +
  theme_bw(base_size = 16)

^{Created on 2021-06-18 by the reprex package (v0.3.0)}

Andrzej · June 18, 2021, 7:21pm

HI @FJCC ,
can you share the data please ?

pfarelo · June 18, 2021, 7:48pm

@FJCC

which program did you use to run? when i run using Rstudio the regression lines do not show up only see the dots

FJCC · June 18, 2021, 8:50pm

The data:

structure(list(Variety = c("Akwa", "Akwa", "Akwa", "Akwa", "Anel", 
"Anel", "Anel", "Anel", "Kwarts", "Kwarts", "Kwarts", "Kwarts", 
"Selie", "Selie", "Selie", "Selie"), Dose = c(25L, 50L, 75L, 
100L, 25L, 50L, 75L, 100L, 25L, 50L, 75L, 100L, 25L, 50L, 75L, 
100L), avg_AB = c(13.2, 30.8, 24.8, 43.2, 5, 13.5, 19.2, 33, 
24.8, 30.2, 42, 50, 8.25, 13.8, 20, 33.5), se = c(5.07, 6.98, 
4.9, 5.93, 3.61, 4.19, 2.72, 2.7, 4.9, 2.76, 4.33, 3.25, 3.67, 
2.52, 2.24, 1.84)), class = "data.frame", row.names = c(NA, -16L
))

FJCC · June 18, 2021, 8:54pm

The code I posted above was run in RStudio. You can just substitute the structure() call that I just posted for the read.csv() call in my code.

Edit: Notice that I changed the formula within geom_smooth and stat_poly_eq.

pfarelo · June 18, 2021, 11:07pm

@FJCC

The problem i am now experiencing is that the graph not showing regression lines but plotted points. Only when i change the Dose to integer or numeric then i get the regression line on the graph. Changing Dose to integer nor numeric changes the values to 1,2,3, and 4 which is not the actual presentation of the graph ...values required are the 25,75 as in your constructed graph

FJCC · June 18, 2021, 11:49pm

Please post the output of

dput(MeanSE_AB)

Put a line with three back ticks before and after your pasted output, like this
```
Your output
```

pfarelo · June 19, 2021, 3:45am

structure(list(Variety = structure(c(1L, 1L, 1L, 1L, 2L, 2L, 
2L, 2L, 3L, 3L, 3L, 3L, 4L, 4L, 4L, 4L), .Label = c("Akwa", "Anel", 
"Kwarts", "Selie"), class = "factor"), Dose = structure(c(1L, 
2L, 3L, 4L, 1L, 2L, 3L, 4L, 1L, 2L, 3L, 4L, 1L, 2L, 3L, 4L), .Label = c("25", 
"50", "75", "100"), class = "factor"), avg_AB = c(13.25, 30.75, 
24.75, 43.25, 5, 13.5, 19.25, 33, 24.75, 30.25, 42, 50, 8.25, 
13.75, 20, 33.5), se = c(5.07004508292157, 6.97892234824669, 
4.89806813229169, 5.9273880058107, 3.60555127546399, 4.18756663059996, 
2.72390213795262, 2.69920623252731, 4.89806813229169, 2.76295649104878, 
4.32600112277906, 3.25137333621173, 3.67301938853737, 2.51956628920818, 
2.23606797749979, 1.84197099403252)), row.names = c(NA, -16L), groups = structure(list(
    Variety = structure(1:4, .Label = c("Akwa", "Anel", "Kwarts", 
    "Selie"), class = "factor"), .rows = structure(list(1:4, 
        5:8, 9:12, 13:16), ptype = integer(0), class = c("vctrs_list_of", 
    "vctrs_vctr", "list"))), row.names = c(NA, -4L), class = c("tbl_df", 
"tbl", "data.frame"), .drop = TRUE), class = c("grouped_df", 
"tbl_df", "tbl", "data.frame"))

FJCC · June 19, 2021, 4:25am

The Dose column in your data is a factor. You can see this with

class(MeanSE_AB$Dose)

Try running the following code before graphing.

MeanSE_AB$Dose <- as.numeric(as.character(MeanSE_AB$Dose))

pfarelo · June 19, 2021, 4:36am

thanks a lot t its working now
i had set dose as : "as.numeric" Not "as.numeric(as.character)"
much appreciated

pfarelo · June 23, 2021, 1:23pm

Hi,

do you maybe perhaps know how one can get the points to start at zero or pass through zero point? I would like to reference that to the zero point

Thanks

FJCC · June 23, 2021, 2:36pm

You can force the fits to go through zero by adding + 0 to the fit formulas.

ggplot(data = MeanSE_AB, aes(x = Dose, y = avg_AB)) +
  geom_point() +
  facet_wrap(~Variety)+
  geom_smooth(method = "lm", se = FALSE, formula = y ~ x + 0) +
  stat_poly_eq(aes(label = paste0("atop(", ..eq.label.., ",", ..rr.label.., ")")),
               formula = y ~ x + 0,
               parse = TRUE) +
  theme_bw(base_size = 16)

system · June 30, 2021, 2:36pm

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.