Thursday, September 14, 2006

Counting

I wanted to know what the solution time is in relation to the times I have seen the problem before:


#times seen before
#problems
avg seconds
median seconds
0
5
11.0
8.6
11511.4
7.9
2157.5
5.1
3118.1
6.2
4
5
6.9
5.0
5
3
8.9
6.1
6
1
3.1
3.1
7
1
2.3
2.3

The selection is a bit wee (56), but it is a lot of work and there are other things to do.
The first 40,000 repetitions of 64,000 are not taken into account, since the history of CTS is somewhat short for unknown reasons.

17 comments:

  1. That's a pretty strong (negative) correlation: -.9 correlation coefficient.

    ReplyDelete
  2. What is the exact reason why you are faster with problems seen before? Do you spot all relevant patterns faster or do you just remember the move and do it without spotting all the relevant patterns?

    ReplyDelete
  3. Mouse,
    that isn't easy to answer. In order to answer the question I analysed a set of 80 correct answers.
    I recognized 3 categories:

    A. I constructed the answer. Allthough you can't construct answers without pattern recognition, the common characteristic is that it is SLOW.

    B. I remembered the moves. Most of the time this is FAST, but sometimes it takes considerable time (20% of the cases)

    C. I recognized the relevant patterns. This is always FAST.

    It isn't always clear cut. Sometimes you construct the first two moves and remembers the third. How slow must pattern recognition be before you call it a construction? etc.

    The distribution was A 30, B 24, C 26. So for the fast answers it is about 50/50 B or C.

    I wouldn't be too surprised if C is B while you forgot that you learned it.

    ReplyDelete
  4. Sounds reasonable and reflects more or less my own experience. If it comes to B, you have to decide whether you want to optimize your rating (playing a tempo the recalled move) or the training effect (spotting the patterns first, irrespective that you know already what to move). I admit that I always take gratefully the cheap point, but I am aware that this is basically a waste of time. You still may spot the pattern after the move, but this is an extremely bad habit which should never be practised.

    ReplyDelete
  5. This comment has been removed by a blog administrator.

    ReplyDelete
  6. Did I ever tell you guys that you're crazy? I'm sure I did! Well, I might have forgotten so here goes: You guys, you're crazy!

    Puh, now it feels a whole lot better. Now I just need to touch the screen 47 times or the world will come to an end. 1...2...3...4...

    ReplyDelete
  7. Tempo, do you have data to do a relation of success/failure vs. #of times you've seen the problem ? Of couse time is important, but may be success is more important than time ? may be, may be not.

    ReplyDelete
  8. Shak, I don't know if I understand you well, but my post of wednesday was about failed problems and the #times seen before. Does that answer your question?

    ReplyDelete
  9. Yes, that's it.

    Another question. How are you keeping your statistics ? I mean, CTS site does not show you anything but the last session, best and worst efforts. Are you copying your session data after doing them ? Are you taking notes for each problem, like "familiar", "unfamiliar", etc ?

    ReplyDelete
  10. Shak, indeed, I copy the session data when I want to research something. For the table in this post I had to go thru the history of each problem and count my name. A daunting task.

    ReplyDelete
  11. How are you copying the session table ? You should be using the following procedure that I found good for getting a SQL database of the problems done.

    1) Select all the text of the table session with your mouse.
    2) Press Ctrl-C, or right click your mouse and select "Copy"
    3) Open an MS Excel file.
    4) Select a cell where you want the data to be copied.
    5) Press Ctrl-V (or right click your mouse and select Paste).

    This is basically a classical Copy and Paste, but some people doesn't know that excel recognizes HTML format.

    Keep your data on one excel file, adding each session below the last row. I'm going to do an excel template sheet so you can obtain the statistics, about time, failure, problem counting, etc.

    Let me know your opinion about it.

    ReplyDelete
  12. Shak, I go to "view source" in my browser and copy and paste it into a html-editor. Does excel copy everything to 1 cell or multiple-cells (which would be very nice)?

    ReplyDelete
  13. OK, now use Ctrl-H to replace " secs" word after times with blank so you can do operations with the time cells.

    Use a formula like this
    =COUNTIF(D2:D16, "failed to solve")

    to count the failed ones

    Use a formula like this

    =COUNTIF($E$2:$E$16,E7)

    in all cells of a new column called for example "#times problem" to get the number of times the problem exist in a given range. E7 is the cell where the problem is.

    Hope you are familiar with excel, if not, please let me know what could you do.

    ReplyDelete
  14. This comment has been removed by a blog administrator.

    ReplyDelete
  15. I can do most things I want in Excel. What I want at the moment is to make an exact table like in this post, but based on a bigger problemset. Say 300 in stead of 56 problems For this post I gathered the information by hand. Do you have an idea to automate (parts of) it?

    ReplyDelete
  16. Sure !,'I've sent you an eMail with an Excel example on that. Don't know how to share a file in the blog.

    ReplyDelete