Skip to main content

Table 7 The results of different dimensions on medical code classification between our method and Word2Vec

From: Refining electronic medical records representation in manifold subspace

Dimension Metric Word2Vec Ours
100 Pearson 69.2 68.8
100 Spearman 63.8 63.5
200 Pearson 69.2 70.1
200 Spearman 63.8 64.3
250 Pearson 69.2 71.2
250 Spearman 63.8 65.6
300 Pearson 69.2 70.8
300 Spearman 63.8 67.3
  1. Bold values represent the best result for each row of data(%). (Original space dimension is 300d,(window start \(\in\) [0,1000], number of MLLE local neighbors = 500, manifold dimensionality = space dimensionality)