: automated levene test and other tests for variable datasets

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

: automated levene test and other tests for variable datasets

Joachim Audenaert
Hello all,

I am writing a script for statistical comparison of means. I'm doing many
field trials with plants, where we have to compare the efficacy of
different treatments on, different groups of plants. Therefore I would
like to automate this script so it can be used for different datasets of
different experiments (which will have different dimensions). An example
dataset is given here under, I would like to compare if the data of 5
columns (A,B,C,D,E) are statistically different from each other, where A,
B, C, D and A are different treatments of my plants and I have 5
replications for this experiment

dataset <- structure(list(A = c(62, 55, 57, 103, 59), B = c(36, 24, 61,
19, 79), C = c(33, 97, 54, 48, 166), D = c(106, 82, 116, 85, 94), E =
c(32, 16, 9, 7, 46)), .Names = c("A", "B", "C", "D",    "E"), row.names =
c(NA, 5L), class = "data.frame")

1) First I would like to do a levene test to check the equality of
variances of my datasets. Currently I do this as follows:

library("car")
attach(dataset)
y <- c(A,B,C,D,E)
group <- as.factor(c(rep(1, length(A)), rep(2, length(B)),rep(3,
length(C)), rep(4, length(D)),rep(5, length(E))))
leveneTest(y, group)

Is there a way to automate this for all types of datasets, so that I can
use the same script for a datasets with any number of columns of data to
compare? My above script only works for a dataset with 5 columns to
compare

2) For my boxplots I use

boxplot(dataset)

which gives me all the boxplots of each dataset, so this is how I want it

3) To check normality I currently use the kolmogorov smirnov test as
follows

ks.test(A,pnorm)
ks.test(B,pnorm)
ks.test(C,pnorm)
ks.test(D,pnorm)
ks.test(E,pnorm)

Is there a way to replace the A, B, C, ... on the five lines into one line
of entry so that the kolmogorov smirnov test is done on all columns of my
dataset at once?

4) if data is normally distributed and the variances are equal I want to
do a t-test and do pairwise comparison, currently like this

pairwise.t.test(y,group,p.adjust.method = "none")

if data is not normally distributed or variances are unequal I do a
pairwise comparison with the wilcoxon test

pairwise.wilcox.test(y,group,p.adjust.method = "none")

But again I would like to make this easier, is there a way to replace the
y and group in my datalineby something so it works for any size of
dataset?

5) Once I have my paiwise comparison results I know which groups are
statistically different from others, so I can add a and b and c to
different groups in my graph. Currently I do this on a sheet of paper by
comparing them one by one. Is there also a way to automate this? So R
gives me for example something like this

A: a
B: a
C: b
D: ab
E: c

All help and commentys are welcome. I'm quite new to R and not a
statistical genious, so if I'm overseeing things or thinking in a wrong
way please let me know how I can improve my way of working. In short I
would like to build a script that can compare the means of different
groups of data and check if they are statistically diiferent

Met vriendelijke groeten - With kind regards,

Joachim Audenaert
onderzoeker gewasbescherming - crop protection researcher

PCS | proefcentrum voor sierteelt - ornamental plant research

Schaessestraat 18, 9070 Destelbergen, Belgi�
T: +32 (0)9 353 94 71 | F: +32 (0)9 353 94 95
E: [hidden email] | W: www.pcsierteelt.be  

Heb je je individuele begeleiding bemesting (CVBB) al aangevraagd? | Het
PCS op LinkedIn
Disclaimer | Please consider the environment before printing. Think green,
keep it on the screen!
        [[alternative HTML version deleted]]


______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: : automated levene test and other tests for variable datasets

Bert Gunter
Sounds like you need to do some of your own "homework"... In
particular, you need to learn how to write your own functions in R to
carry out such tasks.

There are many good tutorials on how to program in R, e.g. the Intro
to R that ships with R and many others that can be found by searching.
Choose one that suits your tastes/learning style and have at it!

Cheers,
Bert

Bert Gunter
Genentech Nonclinical Biostatistics
(650) 467-7374

"Data is not information. Information is not knowledge. And knowledge
is certainly not wisdom."
Clifford Stoll




On Tue, Apr 14, 2015 at 1:07 AM, Joachim Audenaert
<[hidden email]> wrote:

> Hello all,
>
> I am writing a script for statistical comparison of means. I'm doing many
> field trials with plants, where we have to compare the efficacy of
> different treatments on, different groups of plants. Therefore I would
> like to automate this script so it can be used for different datasets of
> different experiments (which will have different dimensions). An example
> dataset is given here under, I would like to compare if the data of 5
> columns (A,B,C,D,E) are statistically different from each other, where A,
> B, C, D and A are different treatments of my plants and I have 5
> replications for this experiment
>
> dataset <- structure(list(A = c(62, 55, 57, 103, 59), B = c(36, 24, 61,
> 19, 79), C = c(33, 97, 54, 48, 166), D = c(106, 82, 116, 85, 94), E =
> c(32, 16, 9, 7, 46)), .Names = c("A", "B", "C", "D",    "E"), row.names =
> c(NA, 5L), class = "data.frame")
>
> 1) First I would like to do a levene test to check the equality of
> variances of my datasets. Currently I do this as follows:
>
> library("car")
> attach(dataset)
> y <- c(A,B,C,D,E)
> group <- as.factor(c(rep(1, length(A)), rep(2, length(B)),rep(3,
> length(C)), rep(4, length(D)),rep(5, length(E))))
> leveneTest(y, group)
>
> Is there a way to automate this for all types of datasets, so that I can
> use the same script for a datasets with any number of columns of data to
> compare? My above script only works for a dataset with 5 columns to
> compare
>
> 2) For my boxplots I use
>
> boxplot(dataset)
>
> which gives me all the boxplots of each dataset, so this is how I want it
>
> 3) To check normality I currently use the kolmogorov smirnov test as
> follows
>
> ks.test(A,pnorm)
> ks.test(B,pnorm)
> ks.test(C,pnorm)
> ks.test(D,pnorm)
> ks.test(E,pnorm)
>
> Is there a way to replace the A, B, C, ... on the five lines into one line
> of entry so that the kolmogorov smirnov test is done on all columns of my
> dataset at once?
>
> 4) if data is normally distributed and the variances are equal I want to
> do a t-test and do pairwise comparison, currently like this
>
> pairwise.t.test(y,group,p.adjust.method = "none")
>
> if data is not normally distributed or variances are unequal I do a
> pairwise comparison with the wilcoxon test
>
> pairwise.wilcox.test(y,group,p.adjust.method = "none")
>
> But again I would like to make this easier, is there a way to replace the
> y and group in my datalineby something so it works for any size of
> dataset?
>
> 5) Once I have my paiwise comparison results I know which groups are
> statistically different from others, so I can add a and b and c to
> different groups in my graph. Currently I do this on a sheet of paper by
> comparing them one by one. Is there also a way to automate this? So R
> gives me for example something like this
>
> A: a
> B: a
> C: b
> D: ab
> E: c
>
> All help and commentys are welcome. I'm quite new to R and not a
> statistical genious, so if I'm overseeing things or thinking in a wrong
> way please let me know how I can improve my way of working. In short I
> would like to build a script that can compare the means of different
> groups of data and check if they are statistically diiferent
>
> Met vriendelijke groeten - With kind regards,
>
> Joachim Audenaert
> onderzoeker gewasbescherming - crop protection researcher
>
> PCS | proefcentrum voor sierteelt - ornamental plant research
>
> Schaessestraat 18, 9070 Destelbergen, België
> T: +32 (0)9 353 94 71 | F: +32 (0)9 353 94 95
> E: [hidden email] | W: www.pcsierteelt.be
>
> Heb je je individuele begeleiding bemesting (CVBB) al aangevraagd? | Het
> PCS op LinkedIn
> Disclaimer | Please consider the environment before printing. Think green,
> keep it on the screen!
>         [[alternative HTML version deleted]]
>
>
> ______________________________________________
> [hidden email] mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: : automated levene test and other tests for variable datasets

Michael Dewey-3
In reply to this post by Joachim Audenaert
You ask quite a lot of questions, I have given some hints about your
first example inline

On 14/04/2015 09:07, Joachim Audenaert wrote:

> Hello all,
>
> I am writing a script for statistical comparison of means. I'm doing many
> field trials with plants, where we have to compare the efficacy of
> different treatments on, different groups of plants. Therefore I would
> like to automate this script so it can be used for different datasets of
> different experiments (which will have different dimensions). An example
> dataset is given here under, I would like to compare if the data of 5
> columns (A,B,C,D,E) are statistically different from each other, where A,
> B, C, D and A are different treatments of my plants and I have 5
> replications for this experiment
>
> dataset <- structure(list(A = c(62, 55, 57, 103, 59), B = c(36, 24, 61,
> 19, 79), C = c(33, 97, 54, 48, 166), D = c(106, 82, 116, 85, 94), E =
> c(32, 16, 9, 7, 46)), .Names = c("A", "B", "C", "D",    "E"), row.names =
> c(NA, 5L), class = "data.frame")
>
> 1) First I would like to do a levene test to check the equality of
> variances of my datasets. Currently I do this as follows:
>
> library("car")
> attach(dataset)
Usually best to avoid this and use the data=parameter or with or within

> y <- c(A,B,C,D,E)
you could use unlist( ) here
> group <- as.factor(c(rep(1, length(A)), rep(2, length(B)),rep(3,
> length(C)), rep(4, length(D)),rep(5, length(E))))
you can get the lengths which you need with
lengtha <- lapply(dataset, length)
or
lengths <- sapply(dataset, length)
depending

then
rep(letters[1:length(lengths)], lengths)
should get you the group variable you want.


I have just typed all those in so there may be typos but at least you
know where to look. I am not suggesting that I think automating all
statistical analyses is necessarily a good idea either.

> leveneTest(y, group)
>
> Is there a way to automate this for all types of datasets, so that I can
> use the same script for a datasets with any number of columns of data to
> compare? My above script only works for a dataset with 5 columns to
> compare
>
> 2) For my boxplots I use
>
> boxplot(dataset)
>
> which gives me all the boxplots of each dataset, so this is how I want it
>
> 3) To check normality I currently use the kolmogorov smirnov test as
> follows
>
> ks.test(A,pnorm)
> ks.test(B,pnorm)
> ks.test(C,pnorm)
> ks.test(D,pnorm)
> ks.test(E,pnorm)
>
> Is there a way to replace the A, B, C, ... on the five lines into one line
> of entry so that the kolmogorov smirnov test is done on all columns of my
> dataset at once?
>
> 4) if data is normally distributed and the variances are equal I want to
> do a t-test and do pairwise comparison, currently like this
>
> pairwise.t.test(y,group,p.adjust.method = "none")
>
> if data is not normally distributed or variances are unequal I do a
> pairwise comparison with the wilcoxon test
>
> pairwise.wilcox.test(y,group,p.adjust.method = "none")
>
> But again I would like to make this easier, is there a way to replace the
> y and group in my datalineby something so it works for any size of
> dataset?
>
> 5) Once I have my paiwise comparison results I know which groups are
> statistically different from others, so I can add a and b and c to
> different groups in my graph. Currently I do this on a sheet of paper by
> comparing them one by one. Is there also a way to automate this? So R
> gives me for example something like this
>
> A: a
> B: a
> C: b
> D: ab
> E: c
>
> All help and commentys are welcome. I'm quite new to R and not a
> statistical genious, so if I'm overseeing things or thinking in a wrong
> way please let me know how I can improve my way of working. In short I
> would like to build a script that can compare the means of different
> groups of data and check if they are statistically diiferent
>
> Met vriendelijke groeten - With kind regards,
>
> Joachim Audenaert
> onderzoeker gewasbescherming - crop protection researcher
>
> PCS | proefcentrum voor sierteelt - ornamental plant research
>
> Schaessestraat 18, 9070 Destelbergen, Belgi�
> T: +32 (0)9 353 94 71 | F: +32 (0)9 353 94 95
> E: [hidden email] | W: www.pcsierteelt.be
>
> Heb je je individuele begeleiding bemesting (CVBB) al aangevraagd? | Het
> PCS op LinkedIn
> Disclaimer | Please consider the environment before printing. Think green,
> keep it on the screen!
> [[alternative HTML version deleted]]
>
>
>
> ______________________________________________
> [hidden email] mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

--
Michael
http://www.dewey.myzen.co.uk/home.html

______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: : automated levene test and other tests for variable datasets

Thierry Onkelinx
In reply to this post by Joachim Audenaert
Dear Joachim,

Storing your data in a long format will make this a lot easier.

library(reshape2)
long.data <- melt(dataset, measure.var = c("A", "B", "C", "D", "E"))
library(car)
leveneTest(value ~ variable, data = long.data)

library(plyr)
ddply(long.data, "variable", function(x){ks.test(x$value})

Best regards,



ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature and
Forest
team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
Kliniekstraat 25
1070 Anderlecht
Belgium

To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to say
what the experiment died of. ~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data. ~ Roger Brinner
The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of data.
~ John Tukey

2015-04-14 10:07 GMT+02:00 Joachim Audenaert <
[hidden email]>:

> Hello all,
>
> I am writing a script for statistical comparison of means. I'm doing many
> field trials with plants, where we have to compare the efficacy of
> different treatments on, different groups of plants. Therefore I would
> like to automate this script so it can be used for different datasets of
> different experiments (which will have different dimensions). An example
> dataset is given here under, I would like to compare if the data of 5
> columns (A,B,C,D,E) are statistically different from each other, where A,
> B, C, D and A are different treatments of my plants and I have 5
> replications for this experiment
>
> dataset <- structure(list(A = c(62, 55, 57, 103, 59), B = c(36, 24, 61,
> 19, 79), C = c(33, 97, 54, 48, 166), D = c(106, 82, 116, 85, 94), E =
> c(32, 16, 9, 7, 46)), .Names = c("A", "B", "C", "D",    "E"), row.names =
> c(NA, 5L), class = "data.frame")
>
> 1) First I would like to do a levene test to check the equality of
> variances of my datasets. Currently I do this as follows:
>
> library("car")
> attach(dataset)
> y <- c(A,B,C,D,E)
> group <- as.factor(c(rep(1, length(A)), rep(2, length(B)),rep(3,
> length(C)), rep(4, length(D)),rep(5, length(E))))
> leveneTest(y, group)
>
> Is there a way to automate this for all types of datasets, so that I can
> use the same script for a datasets with any number of columns of data to
> compare? My above script only works for a dataset with 5 columns to
> compare
>
> 2) For my boxplots I use
>
> boxplot(dataset)
>
> which gives me all the boxplots of each dataset, so this is how I want it
>
> 3) To check normality I currently use the kolmogorov smirnov test as
> follows
>
> ks.test(A,pnorm)
> ks.test(B,pnorm)
> ks.test(C,pnorm)
> ks.test(D,pnorm)
> ks.test(E,pnorm)
>
> Is there a way to replace the A, B, C, ... on the five lines into one line
> of entry so that the kolmogorov smirnov test is done on all columns of my
> dataset at once?
>
> 4) if data is normally distributed and the variances are equal I want to
> do a t-test and do pairwise comparison, currently like this
>
> pairwise.t.test(y,group,p.adjust.method = "none")
>
> if data is not normally distributed or variances are unequal I do a
> pairwise comparison with the wilcoxon test
>
> pairwise.wilcox.test(y,group,p.adjust.method = "none")
>
> But again I would like to make this easier, is there a way to replace the
> y and group in my datalineby something so it works for any size of
> dataset?
>
> 5) Once I have my paiwise comparison results I know which groups are
> statistically different from others, so I can add a and b and c to
> different groups in my graph. Currently I do this on a sheet of paper by
> comparing them one by one. Is there also a way to automate this? So R
> gives me for example something like this
>
> A: a
> B: a
> C: b
> D: ab
> E: c
>
> All help and commentys are welcome. I'm quite new to R and not a
> statistical genious, so if I'm overseeing things or thinking in a wrong
> way please let me know how I can improve my way of working. In short I
> would like to build a script that can compare the means of different
> groups of data and check if they are statistically diiferent
>
> Met vriendelijke groeten - With kind regards,
>
> Joachim Audenaert
> onderzoeker gewasbescherming - crop protection researcher
>
> PCS | proefcentrum voor sierteelt - ornamental plant research
>
> Schaessestraat 18, 9070 Destelbergen, België
> T: +32 (0)9 353 94 71 | F: +32 (0)9 353 94 95
> E: [hidden email] | W: www.pcsierteelt.be
>
> Heb je je individuele begeleiding bemesting (CVBB) al aangevraagd? | Het
> PCS op LinkedIn
> Disclaimer | Please consider the environment before printing. Think green,
> keep it on the screen!
>         [[alternative HTML version deleted]]
>
>
> ______________________________________________
> [hidden email] mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

        [[alternative HTML version deleted]]

______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: : automated levene test and other tests for variable datasets

Joachim Audenaert
Thank you very much for the reply Thierry,

It was very useful for me, currently I updated my script as follows, to be
able to use the same script for different datasets:

adapting my dataset : y <- melt(dataset, na.rm=TRUE) where "na.rm = true"
ommits missing data points

variable <- y[,1]
value <- y[,2]

and then for the tests

leveneTest(value~variable,y)
apply(dataset,MARGIN=2,FUN=function(x) ks.test(x,pnorm)$p.value)

pairwise.t.test(value,variable,p.adjust.method = "none")
pairwise.wilcox.test(value,variable,p.adjust.method = "none")

Met vriendelijke groeten - With kind regards,

Joachim Audenaert
onderzoeker gewasbescherming - crop protection researcher

PCS | proefcentrum voor sierteelt - ornamental plant research

Schaessestraat 18, 9070 Destelbergen, Belgi�
T: +32 (0)9 353 94 71 | F: +32 (0)9 353 94 95
E: [hidden email] | W: www.pcsierteelt.be



From:   Thierry Onkelinx <[hidden email]>
To:     Joachim Audenaert <[hidden email]>
Cc:     "[hidden email]" <[hidden email]>
Date:   15/04/2015 13:31
Subject:        Re: [R] : automated levene test and other tests for
variable datasets



Dear Joachim,

Storing your data in a long format will make this a lot easier.

library(reshape2)
long.data <- melt(dataset, measure.var = c("A", "B", "C", "D", "E"))
library(car)
leveneTest(value ~ variable, data = long.data)

library(plyr)
ddply(long.data, "variable", function(x){ks.test(x$value})

Best regards,



ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature and
Forest
team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
Kliniekstraat 25
1070 Anderlecht
Belgium

To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to
say what the experiment died of. ~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data. ~ Roger Brinner
The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of
data. ~ John Tukey

2015-04-14 10:07 GMT+02:00 Joachim Audenaert <
[hidden email]>:
Hello all,

I am writing a script for statistical comparison of means. I'm doing many
field trials with plants, where we have to compare the efficacy of
different treatments on, different groups of plants. Therefore I would
like to automate this script so it can be used for different datasets of
different experiments (which will have different dimensions). An example
dataset is given here under, I would like to compare if the data of 5
columns (A,B,C,D,E) are statistically different from each other, where A,
B, C, D and A are different treatments of my plants and I have 5
replications for this experiment

dataset <- structure(list(A = c(62, 55, 57, 103, 59), B = c(36, 24, 61,
19, 79), C = c(33, 97, 54, 48, 166), D = c(106, 82, 116, 85, 94), E =
c(32, 16, 9, 7, 46)), .Names = c("A", "B", "C", "D",    "E"), row.names =
c(NA, 5L), class = "data.frame")

1) First I would like to do a levene test to check the equality of
variances of my datasets. Currently I do this as follows:

library("car")
attach(dataset)
y <- c(A,B,C,D,E)
group <- as.factor(c(rep(1, length(A)), rep(2, length(B)),rep(3,
length(C)), rep(4, length(D)),rep(5, length(E))))
leveneTest(y, group)

Is there a way to automate this for all types of datasets, so that I can
use the same script for a datasets with any number of columns of data to
compare? My above script only works for a dataset with 5 columns to
compare

2) For my boxplots I use

boxplot(dataset)

which gives me all the boxplots of each dataset, so this is how I want it

3) To check normality I currently use the kolmogorov smirnov test as
follows

ks.test(A,pnorm)
ks.test(B,pnorm)
ks.test(C,pnorm)
ks.test(D,pnorm)
ks.test(E,pnorm)

Is there a way to replace the A, B, C, ... on the five lines into one line
of entry so that the kolmogorov smirnov test is done on all columns of my
dataset at once?

4) if data is normally distributed and the variances are equal I want to
do a t-test and do pairwise comparison, currently like this

pairwise.t.test(y,group,p.adjust.method = "none")

if data is not normally distributed or variances are unequal I do a
pairwise comparison with the wilcoxon test

pairwise.wilcox.test(y,group,p.adjust.method = "none")

But again I would like to make this easier, is there a way to replace the
y and group in my datalineby something so it works for any size of
dataset?

5) Once I have my paiwise comparison results I know which groups are
statistically different from others, so I can add a and b and c to
different groups in my graph. Currently I do this on a sheet of paper by
comparing them one by one. Is there also a way to automate this? So R
gives me for example something like this

A: a
B: a
C: b
D: ab
E: c

All help and commentys are welcome. I'm quite new to R and not a
statistical genious, so if I'm overseeing things or thinking in a wrong
way please let me know how I can improve my way of working. In short I
would like to build a script that can compare the means of different
groups of data and check if they are statistically diiferent

Met vriendelijke groeten - With kind regards,

Joachim Audenaert
onderzoeker gewasbescherming - crop protection researcher

PCS | proefcentrum voor sierteelt - ornamental plant research

Schaessestraat 18, 9070 Destelbergen, Belgi�
T: +32 (0)9 353 94 71 | F: +32 (0)9 353 94 95
E: [hidden email] | W: www.pcsierteelt.be

Heb je je individuele begeleiding bemesting (CVBB) al aangevraagd? | Het
PCS op LinkedIn
Disclaimer | Please consider the environment before printing. Think green,
keep it on the screen!
        [[alternative HTML version deleted]]


______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.



Heb je je individuele begeleiding bemesting (CVBB) al aangevraagd? | Het
PCS op LinkedIn
Disclaimer | Please consider the environment before printing. Think green,
keep it on the screen!
        [[alternative HTML version deleted]]


______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: : automated levene test and other tests for variable datasets

Joachim Audenaert
In reply to this post by Michael Dewey-3
Hello Michael,

thank you for the reply, it realy helped me to simplify my script.
Basically all my questions are a bit the same, but with your hint I could
solve most of my problems.

Met vriendelijke groeten - With kind regards,

Joachim Audenaert
onderzoeker gewasbescherming - crop protection researcher

PCS | proefcentrum voor sierteelt - ornamental plant research

Schaessestraat 18, 9070 Destelbergen, België
T: +32 (0)9 353 94 71 | F: +32 (0)9 353 94 95
E: [hidden email] | W: www.pcsierteelt.be



From:   Michael Dewey <[hidden email]>
To:     Joachim Audenaert <[hidden email]>,
[hidden email]
Date:   14/04/2015 18:17
Subject:        Re: [R] : automated levene test and other tests for
variable datasets



You ask quite a lot of questions, I have given some hints about your
first example inline

On 14/04/2015 09:07, Joachim Audenaert wrote:
> Hello all,
>
> I am writing a script for statistical comparison of means. I'm doing
many
> field trials with plants, where we have to compare the efficacy of
> different treatments on, different groups of plants. Therefore I would
> like to automate this script so it can be used for different datasets of
> different experiments (which will have different dimensions). An example
> dataset is given here under, I would like to compare if the data of 5
> columns (A,B,C,D,E) are statistically different from each other, where
A,
> B, C, D and A are different treatments of my plants and I have 5
> replications for this experiment
>
> dataset <- structure(list(A = c(62, 55, 57, 103, 59), B = c(36, 24, 61,
> 19, 79), C = c(33, 97, 54, 48, 166), D = c(106, 82, 116, 85, 94), E =
> c(32, 16, 9, 7, 46)), .Names = c("A", "B", "C", "D",    "E"), row.names
=
> c(NA, 5L), class = "data.frame")
>
> 1) First I would like to do a levene test to check the equality of
> variances of my datasets. Currently I do this as follows:
>
> library("car")
> attach(dataset)
Usually best to avoid this and use the data=parameter or with or within

> y <- c(A,B,C,D,E)
you could use unlist( ) here
> group <- as.factor(c(rep(1, length(A)), rep(2, length(B)),rep(3,
> length(C)), rep(4, length(D)),rep(5, length(E))))
you can get the lengths which you need with
lengtha <- lapply(dataset, length)
or
lengths <- sapply(dataset, length)
depending

then
rep(letters[1:length(lengths)], lengths)
should get you the group variable you want.


I have just typed all those in so there may be typos but at least you
know where to look. I am not suggesting that I think automating all
statistical analyses is necessarily a good idea either.

> leveneTest(y, group)
>
> Is there a way to automate this for all types of datasets, so that I can
> use the same script for a datasets with any number of columns of data to
> compare? My above script only works for a dataset with 5 columns to
> compare
>
> 2) For my boxplots I use
>
> boxplot(dataset)
>
> which gives me all the boxplots of each dataset, so this is how I want
it

>
> 3) To check normality I currently use the kolmogorov smirnov test as
> follows
>
> ks.test(A,pnorm)
> ks.test(B,pnorm)
> ks.test(C,pnorm)
> ks.test(D,pnorm)
> ks.test(E,pnorm)
>
> Is there a way to replace the A, B, C, ... on the five lines into one
line
> of entry so that the kolmogorov smirnov test is done on all columns of
my

> dataset at once?
>
> 4) if data is normally distributed and the variances are equal I want to
> do a t-test and do pairwise comparison, currently like this
>
> pairwise.t.test(y,group,p.adjust.method = "none")
>
> if data is not normally distributed or variances are unequal I do a
> pairwise comparison with the wilcoxon test
>
> pairwise.wilcox.test(y,group,p.adjust.method = "none")
>
> But again I would like to make this easier, is there a way to replace
the

> y and group in my datalineby something so it works for any size of
> dataset?
>
> 5) Once I have my paiwise comparison results I know which groups are
> statistically different from others, so I can add a and b and c to
> different groups in my graph. Currently I do this on a sheet of paper by
> comparing them one by one. Is there also a way to automate this? So R
> gives me for example something like this
>
> A: a
> B: a
> C: b
> D: ab
> E: c
>
> All help and commentys are welcome. I'm quite new to R and not a
> statistical genious, so if I'm overseeing things or thinking in a wrong
> way please let me know how I can improve my way of working. In short I
> would like to build a script that can compare the means of different
> groups of data and check if they are statistically diiferent
>
> Met vriendelijke groeten - With kind regards,
>
> Joachim Audenaert
> onderzoeker gewasbescherming - crop protection researcher
>
> PCS | proefcentrum voor sierteelt - ornamental plant research
>
> Schaessestraat 18, 9070 Destelbergen, Belgi�
> T: +32 (0)9 353 94 71 | F: +32 (0)9 353 94 95
> E: [hidden email] | W: www.pcsierteelt.be
>
> Heb je je individuele begeleiding bemesting (CVBB) al aangevraagd? | Het
> PCS op LinkedIn
> Disclaimer | Please consider the environment before printing. Think
green,
> keep it on the screen!
>                [[alternative HTML version deleted]]
>
>
>
> ______________________________________________
> [hidden email] mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

--
Michael
http://www.dewey.myzen.co.uk/home.html



Heb je je individuele begeleiding bemesting (CVBB) al aangevraagd? | Het
PCS op LinkedIn
Disclaimer | Please consider the environment before printing. Think green,
keep it on the screen!

        [[alternative HTML version deleted]]

______________________________________________
[hidden email] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.