turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- Does anybody know which algorithm JMP uses to crea...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 18, 2014 2:35 AM
(2781 views)

It seems to be pivot-by-middle proposed by Ben Shneiderman, but I'm not sure. Does anybody know this?

Thanks!

Solved! Go to Solution.

1 ACCEPTED SOLUTION

Accepted Solutions

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 18, 2014 1:44 PM
(4745 views)

It actually is a variation on Split. It recursively divides the space, splitting the rectangle in the larger dimension. The variation is that it doesn't split in half but rather it uses the golden ratio. This algorithm preserves the order of the data but it does result in some rectangles with a larger aspect ratio.

We are looking at new tiling algorithms for a future version of JMP.

4 REPLIES

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 18, 2014 1:44 PM
(4746 views)

It actually is a variation on Split. It recursively divides the space, splitting the rectangle in the larger dimension. The variation is that it doesn't split in half but rather it uses the golden ratio. This algorithm preserves the order of the data but it does result in some rectangles with a larger aspect ratio.

We are looking at new tiling algorithms for a future version of JMP.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 22, 2014 12:40 AM
(2600 views)

Thank you. I'm trying to reproduce the treemap produced by JMP. While the treemap produced by me and that by JMP are similar overall, there are some difference.

Let me show you an example. I used the first twenty rows in Cities.jmp which is used by JMP to show the feature of treemap.

index city POP

1 | ALBANY | 846 |

2 | ALBUQUERQUE | 486 |

3 | ATLANTA | 2657 |

4 | ATLANTIC CITY | 303 |

5 | BALTIMORE | 2303 |

6 | BOSTON | 2842 |

7 | BURLINGTON | 127 |

8 | CHARLESTON | 502 |

9 | CHARLOTTE | 1091 |

10 | CHEYENNE | 76 |

11 | CHICAGO | 6199 |

12 | CINCINNATI | 1438 |

13 | CLEVELAND | 1851 |

14 | DENVER | 1645 |

15 | DES MOINES | 385 |

16 | DETROIT | 4362 |

17 | DUBUQUE | 91 |

18 | GALVESTON-T.C. | 211 |

19 | HARRISBURG | 584 |

20 | HARTFORD | 748 |

All POP sum to 28747, the first nine POP sum to 11157. The ratio is 11157/28747=0.3881, it already exceeded the golden ratio 0.382. However, the treemap produced by JMP further included the tenth datum in the upper part. So my question is that whether JMP uses exactly 0.382 as the split ratio or apart from this ratio, there are other factors affecting which part the datum should belong to(eg. to optimize the aspect ratio).

The two pictures below are produced by JMP and me, respectively. Notice the rectangles in the red area represent the same datum, but in different position. The layout box are both square.

Thanks

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 22, 2014 6:20 AM
(2600 views)

The difference is in the aspect ratio we are using. Looking at the code, we are using .39 instead of .382. When I change the code to use .382, I am able to reproduce your treemap. If you use .39 you should be able to reproduce the treemap in JMP.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 22, 2014 9:43 PM
(2600 views)

Yes, when I use 0.39, the results are the same. Thanks a lot!