3 October 2014

Here is an 'indication' of how good fares this method.
For 2010 and for >5R in the Greek Region, I had 5 correct predictions with less than a day window. This is like saying exact dates. Window 1. I had 50% success rate. The probability of those 5 successes is less than 2%.
In the same year with a window 2 (ie 1 day off) I had 7 successes, 70% success rate.
The probability of  7 successes is now only 2.5% 

My method therefore has promise.

22 comments:

  1. It looks like you're figuring the chances of getting 5 hits by multiplying the chances of a hit by itself 5 times.

    That's not the way to do it. You should use the z-binomial equations instead.

    Roger

    ReplyDelete
  2. Thank you for your contribution Roger. The correct comment I suggest is.....Dear Anthony, in trying to read between the lines I have a feeling that you might be multiplying etc etc The normal way to do it is to apply a binomial instead if you are not doing it already. ?..etc etc ..and you end up with a positive note to encourage research of new ideas so that the earthquake prediction area which is in mess because some retired people have made no significant progress but only to be able to say there will be an earthquake within 10 years of level so and so....what a dire situation....and instead of encouraging young people with new ideas and be positive, after all most new ideas are not correct, but one day something might come up to get you guys out of the sh..t and to be able soon to stand up and give an earthquake report like a weather report on TV. Instead all you get is stifling from you and crap conclusions based on ,scientific, killing of new ideas.......By the way I am looking for enthusiastic research students. ....wink Enjoy golf and be happy.

    ReplyDelete
  3. Anthony;

    I took your table 1 dates and compared them to the Centennial Catalog, finding 47 hits.

    Then I ran it again, adding 5 days to each date and got 48 hits.

    Roger

    ReplyDelete
  4. Roger it is not clear to me what you have done. Do you mean you have shifted the dates uniformly by 5?

    ReplyDelete
  5. As you must be fast asleep in the usa, I guess that the answer maybe yes thats what you have done. Well, I mentioned an improvement and there I have done an exhaustive search. Thats what I hope to publish as there are improvements of course. Yesterday I listed 8 new ideas which I have examined but have not yet released. There is so much to do and I hope to get some new researchers to help out on the numbers. Roger may wish to help the research ...wink....but I am not convinced yet about your intensions as golf maybe more important to you also the joy of killing new ideas... (smiles)

    ReplyDelete
  6. Yes, I shifted each date forward by 5 days. In another test, I shifted each date by a random number from 1 to 10 and got 47 hits.

    My goal is to test your method to see if it actually has merit.

    It's like a game; you set them up, I try to knock them down. If I can't knock them down, you're onto something worthwhile.

    That's how science works.

    Roger

    ReplyDelete
  7. Well you use random numbers and try to guess october earthquake dates.

    ReplyDelete
  8. Mag 4 in Greece?

    Who cares.

    But I'd need to know your window size do do it properly

    Roger

    ReplyDelete
  9. Well I am very pleased you have discovered a better method Roger. This is great.
    I post in here in advance Global Earthquakes just like the Greek Region but with >6.-6.5R
    Now on the topic of Grrek Earthquakes I draw a line on 4R because of the regional frequency of earthquake intensity strength. Above 4R is not noise of course.

    ReplyDelete
    Replies
    1. Anthony;

      Since you don't specify window size, I predict every third day starting with Oct 2, window size 3 days.

      Roger

      Delete
  10. Hi Roger
    You said you did checking on odds for a 4+ quake in the Greek Region.
    for 1 day window, its 0.164
    for 2 day window its 0.250
    for 3 day window its 0.389
    based on the ANSS catalogue for 1973-2013.

    Thanks Roger.
    What do you think of the 2010 odds I have posted above at the top?
    I repost it here....
    Here is an 'indication' of how good fares this method.
    For 2010 and for >5R in the Greek Region, I had 5 correct predictions with less than a day window. This is like saying exact dates. Window 1. I had 50% success rate. The probability of those 5 successes is less than 2%.
    In the same year with a window 2 (ie 1 day off) I had 7 successes, 70% success rate.
    The probability of 7 successes is now only 2.5%

    ReplyDelete
    Replies
    1. Anthony;

      Bad statistics. You can't pick out the best hits from a series and claim the odds for just those hits.

      It's like tossing 100 coins and boasting about the 10 heads.

      Roger

      Delete
    2. I m not picking the best. You are not modelling properly the problem

      Delete
  11. Just to kindly remind you that to conclude what you have on Table 1 based on shifting my dates or even adding those random dates to my data is not sufficient. You really need to compare it with completely new random dates not a purturbation of my dates....

    ReplyDelete
    Replies
    1. Anthony;

      Fair enough.

      Ok, I picked 100 dates at random and compared them to the Centennial Catalog, using a 6 day window.

      I got 68 hits where the expected was 63 hits.

      Way better than your 47 hits.

      Roger

      Delete
  12. roger that it NOT what I do. Picking up the best and ignoring the rest? What are you talking about! Lol Anyway my final results seem to me pretty good.

    ReplyDelete
    Replies
    1. Anthony;

      You were talking about the few best hits in the post to which I was responding.

      Roger

      Delete
    2. The best hits yes but the chances are correct.

      Delete
  13. Roger you must admit my method is probably the best there is. I predict while others can't. Proof is I need time to collect my hits while I should be spending more time on my new ideas.

    ReplyDelete