test1.save

c++实现的KNN库:建立高维度的K-d tree,实现K邻域搜索

------------------------------------------------------------
ann_test: Version 1.0 
    Copyright: David M. Mount and Sunil Arya.
    Latest Revision: Mar 1, 2005.
------------------------------------------------------------

validate = on   (Warning: this may slow execution time.)
stats = query_stats
[Read Data Points:
  data_size  = 20
  file_name  = test1-data.pts
  dim        = 2
]
[Read Query Points:
  query_size = 10
  file_name  = test1-query.pts
  dim        = 2
]
[Build ann-structure:
  split_rule    = suggest
  shrink_rule   = none
  data_size     = 20
  dim           = 2
  bucket_size   = 1
  process_time  = 0 sec
  (Structure Statistics:
    n_nodes          = 39 (opt = 40, best if < 400)
        n_leaves     = 20 (0 contain no points)
        n_splits     = 19
        n_shrinks    = 0
    empty_leaves     = 0 percent (best if < 50 percent)
    depth            = 6 (opt = 4, best if < 17)
    avg_aspect_ratio = 1.48847 (best if < 20)
  )
]
(Computing true nearest neighbors for validation.  This may take time.)
[Run Queries:
  query_size    = 10
  dim           = 2
  search_method = standard
  epsilon       = 0
  near_neigh    = 3
  true_nn       = 13
  query_time    = 0 sec/query (biased by perf measurements)
  (Performance stats:  [      mean :    stddev ]<      min ,       max >
    leaf_nodes       = [       6.3 :     2.751 ]<        4 ,        11 >
    splitting_nodes  = [       8.8 :     3.676 ]<        5 ,        15 >
    shrinking_nodes  = [         0 :         0 ]<        0 ,         0 >
    total_nodes      = [      15.1 :      6.35 ]<        9 ,        26 >
    points_visited   = [       6.3 :     2.751 ]<        4 ,        11 >
    coord_hits/pt    = [      0.57 :    0.2201 ]<     0.35 ,      0.95 >
    floating_ops_(K) = [     0.156 :    0.0563 ]<    0.101 ,     0.254 >
    average_error    = [         0 :         0 ]<        0 ,         0 >
    rank_error       = [         0 :         0 ]<        0 ,         0 >
  )
]
[Run Queries:
  query_size    = 10
  dim           = 2
  search_method = priority
  epsilon       = 0
  near_neigh    = 3
  true_nn       = 13
  query_time    = 0 sec/query (biased by perf measurements)
  (Performance stats:  [      mean :    stddev ]<      min ,       max >
    leaf_nodes       = [       5.9 :     2.025 ]<        4 ,         9 >
    splitting_nodes  = [       8.7 :     3.498 ]<        5 ,        15 >
    shrinking_nodes  = [         0 :         0 ]<        0 ,         0 >
    total_nodes      = [      14.6 :      5.42 ]<        9 ,        24 >
    points_visited   = [       5.9 :     2.025 ]<        4 ,         9 >
    coord_hits/pt    = [     0.535 :    0.1667 ]<     0.35 ,       0.8 >
    floating_ops_(K) = [    0.1719 :   0.05861 ]<    0.114 ,     0.267 >
    average_error    = [         0 :         0 ]<        0 ,         0 >
    rank_error       = [         0 :         0 ]<        0 ,         0 >
  )
]