Open bigfatnoob opened 7 years ago
Topic 0 :: Min = -0.05, Median = 0.00, Max = 0.05, IQR = 0.02
Topic 1 :: Min = -0.05, Median = 0.00, Max = 0.03, IQR = 0.00
Topic 2 :: Min = -0.02, Median = 0.00, Max = 0.03, IQR = 0.00
Topic 3 :: Min = -0.02, Median = 0.00, Max = 0.03, IQR = 0.00
Topic 4 :: Min = -0.01, Median = 0.00, Max = 0.02, IQR = 0.00
Topic 5 :: Min = -0.01, Median = 0.00, Max = 0.01, IQR = 0.00
Topic 6 :: Min = -0.03, Median = 0.00, Max = 0.02, IQR = 0.01
Topic 7 :: Min = 0.00, Median = 0.00, Max = 0.02, IQR = 0.00
Topic 8 :: Min = -0.04, Median = 0.00, Max = 0.04, IQR = 0.00
Topic 9 :: Min = -0.01, Median = 0.00, Max = 0.01, IQR = 0.00
Documents are sent back to the trained model and the delta for each topic's contribution in each document is computed. The stats are reported. We can see that the median and IQR is almost 0 in most cases indicating stability of topics.
No of Papers: 23
Log Likelihood for topics
Topics
Heatmap
Clustered Heatmap