2657 Productions News - Latest Comments

Re: Stratified random sampling in R from a data frame

mrdwab — Thu, 19 Jan 2012 12:26:05 -0000

Hi. I hope you have found the function useful.

It sounds like you've solved the problem you were having, but I'm not sure I understand your follow up response here. Do you have an example you can share?

Re: Stratified random sampling in R from a data frame

B Li — Thu, 19 Jan 2012 11:12:32 -0000

I think I've found the reason for that problem. The key thing is the sample() function used in this stratified() function. When there is only one numeric value (say, 10) to be sampled from, the sample() function samples by default from 1 to 10.
Sorry for the misleading message I posted earlier.

Re: Stratified random sampling in R from a data frame

B Li — Thu, 19 Jan 2012 08:33:21 -0000

There seems to be a bug for large data set with lots of groups. In my case, there are 24000 observations and 2089 groups. I used this function to sample one observation from each group. No errors and the total sample size is alway 2089, but there are always some groups that are sampled with 2 observations and some are sampled with 0 observation.
The sample size is very different from group to group in my case. Maybe that's what you mean in the NOTE of the function "NOTE: Not tested on datasets with LOTS of groups or with HUGE differences in group sizes. Probably INCREDIBLY inefficient." ?

Re: An interview with Ananda Mahto

Rinki gupta — Fri, 06 Jan 2012 11:24:14 -0000

Very informative interview Ananda!

Re: Am I inconsistent?

Akhileshwar Singh — Fri, 30 Sep 2011 07:04:44 -0000

Hi Ananda,

Greetings2You

Being a Academy student it is really difficult to understand your motive behind being sarcastic on students when they ask for some software (I can understand this after I came out of academy). I didn't had chance to went through your post earlier but as I went through, I couldn't stop me writing this reply. I (I believe all PDM 10) always had a special respect for you as a teacher and probably the reason for this was you values and energy for teaching student. I can bet that you are the only one in academy who never hesitated to teach (when you are approached by us) us.

There is nothing specific in this reply about the post. It is the memories of past which over shrouded me and led to memorize you through this reply. I would request you to forgive me (my friends) for all know, unknown mistakes I (we) might have done during the "TWO YEARS" period of time.

Yours Sincerely
Akhilesh

Re: Suckerpunch (10″ and split flexi with Born Against)

patrick — Tue, 20 Sep 2011 19:20:53 -0000

i don't think they ever made a east coast tour, and i too didn't have friends with cars. song for karl is one of favorites.

Re: Sometimes, I wish I could turn my editing eyes off

Kunj Mann — Mon, 19 Sep 2011 08:19:07 -0000

Interesting! I must mention that when I went to Nagapattinam, I saw exactly the same scene. For a moment, I wondered if I took this picture? If pictures usually depict a 'point of time' then Nagapattinam beach scene would be an exception. I guess, coming across such pictures may not have taken much time for you, or tell me if it is otherwise.

Re: Quickly reshaping data from “wide” to “long” formats in R

Omaransi — Tue, 19 Jul 2011 11:30:05 -0000

Hey guys,
I was looking for months someone to explain for me this topic, because I was trying to calculate the age range of household characteristic from one single variable.
It is VERY USEFUL topic and no body can know it unless s/he worked on it.

Many thanks for your effort.

Regards
Omar

Re: Using the reshape package in R for pivot-table-like functionality

mrdwab — Fri, 01 Jul 2011 14:35:32 -0000

I still need to get to grips with plyr (pun intended?), but it appears it could also help with the wide-to-long reshape example. That example isn't a simple transpose example though--if it were, the dimensions of the output would be the same as the input, but with the lengths of the rows and columns switched.

Re: Quickly reshaping data from “wide” to “long” formats in R

mrdwab — Fri, 01 Jul 2011 14:27:08 -0000

Thanks!

I know it's not exactly useless knowledge, but honestly speaking, I don't need R or much data organizing/manipulating/analyzing skills in my work... hence the "useless knowledge" part. This is more something I do out of a personal interest to learn and explore things that I find interesting, and hopefully share what I've found in a way that others might find useful.

Re: Using the reshape package in R for pivot-table-like functionality

isomorphisms — Fri, 01 Jul 2011 01:21:34 -0000

I agree, melt/recast/reshape is essentially a better (harder to use but better) version of Pivot tables / OLAP cubes. I wouldn't say your other post was wrong exactly, just not using the full power of plyr.
I think in the other one you could have just transposed the data, no?

Re: Quickly reshaping data from “wide” to “long” formats in R

isomorphisms — Fri, 01 Jul 2011 01:18:02 -0000

I don't think this belongs under "useless knowledge" — it's very useful!

Re: We 2 Ours 1-1

mrdwab — Wed, 22 Jun 2011 12:58:08 -0000

Glad you enjoyed it. I enjoyed making it!

Re: The new sample size calculator for R (already)

mrdwab — Wed, 22 Jun 2011 12:47:16 -0000

You're welcome.

Just out of curiosity, do you use R at your work? I'm curious about R's uptake in India's professional sector.

Re: We 2 Ours 1-1

Shantigupta211087 — Wed, 22 Jun 2011 06:06:40 -0000

It is great Ananda. Nice painting with present message.

Re: The new sample size calculator for R (already)

Prashanth NS — Wed, 22 Jun 2011 02:15:48 -0000

Thanks! Very useful.

Re: Stratified random sampling in R from a data frame

mrdwab — Fri, 20 May 2011 14:56:46 -0000

Of course, standard tricks can also be used. For instance, if you wanted to take a sample of only groups "CA" and "NY" and drop "TX", you can use the following:


> stratified(dat[dat$E!="TX",], 1, 5, .1, 1)
   A  B           C   D  E
2 36 DD  0.33295037 0.2 CA
4 44 CC  0.70021365 0.8 CA
1 23 DD  0.61072635 0.5 NY
3 37 DD  1.06309984 1.5 NY
5 52 EE  0.04211587 1.7 NY
6 91 BB -1.91435943 0.7 NY

Re: New (old) Pictures

mrdwab — Sun, 15 May 2011 09:44:11 -0000

Links have been updated to redirect to the newest photo site on mrdwab.com.... Self-hosting photos has been a pain.

Re: Exciting day!

mrdwab — Sun, 15 May 2011 09:34:09 -0000

Well, the site is gone, but she Amy has moved on to much bigger things at http://tnsms.org. Be sure to check it out!

Re: Using the reshape package in R for pivot-table-like functionality

mrdwab — Sat, 30 Apr 2011 12:08:11 -0000

Mauricio,

I'm not sure I understand entirely what you're looking for. Can you provide an example of the raw data and describe what you want to end up with?

Re: Using the reshape package in R for pivot-table-like functionality

Mauricio — Fri, 29 Apr 2011 11:07:41 -0000

Would be great if you could comment on whether or not 'reshape' can be used to create output similar to that of ftable ... that is, how to create something that looks like a pivot table with categorical variables on both the rows and columns ... but with values based on functions (i.e. sum)

Re: Making an A6 booklet in OpenOffice.org the easy way

brookmounts — Mon, 25 Apr 2011 18:53:30 -0000

Wow, this one is hugely informative post about making booklet. And the whole post really shares the importance regarding information. I am really totally crazy about this post.

booklet printing

Re: Using the reshape package in R for pivot-table-like functionality

mrdwab — Wed, 30 Mar 2011 03:00:28 -0000

You're welcome. I think solving problems in R helps generate new ideas anyway...

Re: Using the reshape package in R for pivot-table-like functionality

Edmundchen — Tue, 29 Mar 2011 04:57:29 -0000

Thanks for your sharing! It at least gives me some ideas when difficulties keep coming to me~~

Re: The new sample size calculator for R (already)

mrdwab — Tue, 29 Mar 2011 02:57:38 -0000

Shanti,

The calculation part is useful, but what I like most is being able to see multiple scenarios in a single place. For instance, what would our samples be at different confidence levels? What about different margins of error? Using this function--or slightly modified versions of this function, it is easy to see how changing these values would affect your sample size.