Sample from data

[Stef Appel]

Hello, everybody.
I am a beginner, and need help for my bachelor thesis.
I have 600 numbers and need to get samples from that.
To simplifise, I need to sample by random choice 30 numbers from that 600, then to get a mean from that 30.
This operation I need to do 100 times. So, in the end I should have 100 mean-numbers.
What is the best way to do it? can you explain, please, how to do it step by step.

Thanks a lot,
Stefan.

[Carlo Lazzaro]

Stef:
welcome to the list.
What follows is possibly what you're looking for:

. set obs 600
number of observations (_N) was 0, now 600

. g A=4*uniform()

. set seed 1

. g double mean=.
(600 missing values generated)

. quietly forvalues i = 1/100 {
.           preserve
. sample 30 if A !=., count
.          sum A
.          scalar Am = r(mean)
.          restore
.          replace mean = Am in `i'
. }

All credits goes out to Brian Poi, who helped me out some years ago for a similar probem. Obviously, all possible mistakes are my own responsibility..

[Stef Appel]

Hi, Carlo. Thank you very much for aour responce.
I tried to use your code, but I think, I did comething wrong since there is no result. Can you please look at my code?


. import excel "F:\Bachelor\584 st dev.xlsx", sheet("Tabelle1")

. set obs 584
number of observations (_N) was 584, now 584

. set seed 1

. g double mean=.
(584 missing values generated)

. quietly forvalues i= 1/100 {
. preserve
. sample 30 if A !=., count
. sum A
. scalar Am = r(mean)
. restore
. replace mean = Am in `i'
. }

[Stef Appel]

Stef:
welcome to the list.
What follows is possibly what you're looking for:

. set obs 600
number of observations (_N) was 0, now 600

. g A=4*uniform()

. set seed 1

. g double mean=.
(600 missing values generated)

. quietly forvalues i = 1/100 {
. preserve
. sample 30 if A !=., count
. sum A
. scalar Am = r(mean)
. restore
. replace mean = Am in `i'
. }

All credits goes out to Brian Poi, who helped me out some years ago for a similar probem. Obviously, all possible mistakes are my own responsibility..

Hi, Carlo. Thank you very much for aour responce.
I tried to use your code, but I think, I did comething wrong since there is no result. Can you please look at my code?

. import excel "F:\Bachelor\584 st dev.xlsx", sheet("Tabelle1")

. set obs 584
number of observations (_N) was 584, now 584

. set seed 1

. g double mean=.
(584 missing values generated)

. quietly forvalues i= 1/100 {
. preserve
. sample 30 if A !=., count
. sum A
. scalar Am = r(mean)
. restore
. replace mean = Am in `i'
. }




Best regards,
Stef.

[Carlo Lazzaro]

Stef:
works for me.
Hint: copy and paste the code on a do-file, scroll down -Tools- and then select and click on -Execute selection (do)-

set obs 584
g A=4*uniform()
set seed 1
g double mean=.
quietly forvalues i = 1/100 {
preserve
sample 30 if A !=., count
summ A
scalar Am = r(mean)
restore
replace mean = Am in `i'
}

[Stef Appel]

Stef:
works for me.
Hint: copy and paste the code on a do-file, scroll down -Tools- and then select and click on -Execute selection (do)-

set obs 584
g A=4*uniform()
set seed 1
g double mean=.
quietly forvalues i = 1/100 {
preserve
sample 30 if A !=., count
summ A
scalar Am = r(mean)
restore
replace mean = Am in `i'
}

Carlo, it does not work. I send you my printscreen.

[Carlo Lazzaro]

Steff:
try this:

drop _all
set obs 584
g A=4*uniform()
set seed 1
g double mean=.
quietly forvalues i = 1/100 {
preserve
sample 30 if A !=., count
summ A
scalar Am = r(mean)
restore
replace mean = Am in `i'
}

[Stef Appel]

Steff:
try this:

drop _all
set obs 584
g A=4*uniform()
set seed 1
g double mean=.
quietly forvalues i = 1/100 {
preserve
sample 30 if A !=., count
summ A
scalar Am = r(mean)
restore
replace mean = Am in `i'
}

Carlo,
the result, still can't find the numbers:

Best regards,
Stef.

[Carlo Lazzaro]

Stef:
have you tried:

su A mean

[Stef Appel]

Stef:
have you tried:

su A mean

Carlo:
Where should I put it?

[Carlo Lazzaro]

Stef:
try:

drop _all
set obs 584
g A=4*uniform()
set seed 1
g double mean=.
quietly forvalues i = 1/100 {
preserve
sample 30 if A !=., count
summ A
scalar Am = r(mean)
restore
replace mean = Am in `i'
}
su A mean

[Stef Appel]

Stef:
try:

drop _all
set obs 584
g A=4*uniform()
set seed 1
g double mean=.
quietly forvalues i = 1/100 {
preserve
sample 30 if A !=., count
summ A
scalar Am = r(mean)
restore
replace mean = Am in `i'
}
su A mean

Carlo, I got a result. I have two questions. First, I wanted to get 100 numbers. Here I see a mean for of 2, 049231. If I understand right, it is a mean number of 100 numbers, which we wanted to get initially. Is it right? If yes, it is alright. Actually I need this number.
But, I wanted to use my numbers, which I have in my excel sheet. Is it possible? I have 584 numbers, and sample should use them.

Thanks a lot,
Stef.

[Carlo Lazzaro]

Stef:
- you have the overall mean of 100 means, each calculated on a random sample of 30 out 584 observations, as per your request;
- you can import your spreadsheet in Stata via File>Import><select the option which is right for you>

[Stef Appel]

Stef:
- you have the overall mean of 100 means, each calculated on a random sample of 30 out 584 observations, as per your request;
- you can import your spreadsheet in Stata via File>Import><select the option which is right for you>

Carlo:
I already imported my excel sheet. But numbers in my excel sheet vary between 0,005... and 0,063.... I mean, the mean of 2,049231 can not be right for my numbers.
Sorry for your time, but I really need it.

I try to explain again, I have 584 numbers that vary between 0,005... and 0,063...
First, I need a random sample of 30 numbers from 584. Then a mean from that 30 numbers.
This opperation should be done 100 times. In the end, I have 100 numbers, and from them I need to get a mean.

I think, the code is absolutely correct, and refers my needs. But result of 2, 049231 is not right. Do you know what I mean?


Best regards,
Stef.

[Carlo Lazzaro]

Stef:
simply replace your values in A (or import your 584-number variable and name it -A- in Stata.dta) and run the rest of the code.