Pattern recognition in APL with application to reactor diagnostics

Skomorokhov, Alexander O.; Slepov, Michael Timofeevich

doi:10.1145/327600.327643

Cited by 4 publications

(6 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…z~data Predict tree;por;l;];k [1] x(l=ppdata)/'data÷. [.S]data, [2] z~tO [3] :For k :In t+pdata [4] 1~1 [5] :While -tree[i;2]=o [6] (] por)~tree[l;2 3] [7] :If l=.…”

Section: Using the Treeunclassified

“…In the line [4] we call the function Cut to search for trees with minimal costs for all tree sizes. This tree sequence is used later 0rues [6][7][8][9][10][11][12][13][14][15][16]) to produce the final sequence of the cost-complexity optimal trees. The resuh is the 3-columns matrix.…”

Section: K (T') = C(t') + K X Size(t')mentioning

confidence: 99%

“…[0] z~-n CrossValidate tree;data;e;ee ;m;rnd;s; sz; t ; test ; train [1] (tree data)q-tree [2] sz~,l÷[2]Prune tree [3] rnd÷n Samples+pdata [~] z4-0 3pO [5] :For n :In sz [5] e~-to [7] :For m :In rnd [8] test÷data[m;] [9] train÷data [ernd-=m; ] [10]…”

Section: Cross-validauon and Choosing The Final Treementioning

confidence: 99%

“…First, the function prunes the tree (line [2]). Then, for every tree in the obtained sequence (line [5]), the function grows a chssification tree n times (line [7]). Every time a new training set is used (line [10]).…”

Section: Cross-validauon and Choosing The Final Treementioning

confidence: 99%

“…[0] z+class Partitlon x:y;por;p [1] z~to [2] por÷Unlque x [3] :If l=-x [4] por÷por[~por] [5] i(l<ppor)/'por~.5*2+/por' [6] :Else [7] por÷Subsets pot [8] :EndIf [9] :…”

mentioning

confidence: 99%

See 4 more Smart Citations

Classification trees in APL

Skomorokhov

Kutinsky

2000

Proceedings of the 2001 Conference on APL: An Arrays Odyssey

View full text Add to dashboard Cite

This paper considers the problem of classification tree based data analysis. Among the topics discussed in the paper axe: growing a classification tree using CART style exhaustive search for spl/ts, selecting the tight size for the tree using minimal cost-complexity cross-validation pruning, and examples of the application of classification trees.The algorithms are implemented in Dyalog APL.Application example is based on data from vibration monitoring equipment installed on a Nuclear Power Plant in Novovoronehz, Russia and includes classification of vib~tion spectra of steam generators or coolant pumps and classification of vibration spectra of steam generators of different coolant loops.

show abstract

“…z~data Predict tree;por;l;];k [1] x(l=ppdata)/'data÷. [.S]data, [2] z~tO [3] :For k :In t+pdata [4] 1~1 [5] :While -tree[i;2]=o [6] (] por)~tree[l;2 3] [7] :If l=.…”

Section: Using the Treeunclassified

Section: K (T') = C(t') + K X Size(t')mentioning

confidence: 99%

Section: Cross-validauon and Choosing The Final Treementioning

confidence: 99%

Section: Cross-validauon and Choosing The Final Treementioning

confidence: 99%

“…[0] z+class Partitlon x:y;por;p [1] z~to [2] por÷Unlque x [3] :If l=-x [4] por÷por[~por] [5] i(l<ppor)/'por~.5*2+/por' [6] :Else [7] por÷Subsets pot [8] :EndIf [9] :…”

mentioning

confidence: 99%

See 3 more Smart Citations