Relevant Links: Emily Research Home My Journal On Another Note Clustering Methods and Analyses: KNN Clustering K-Median Clustering Kaplan-Meier Estimation Top Genes based on Variance Cross-Validation Significant Genes

Possible Cluster Causes Back to K-Median results

The Harvard Dataset provided statistics on the age, gender, smoking habits, and percentage of the maximum tumor size on the patients within its study. Using these, the average and variance were taken over the members of each group. The results are as follows: Cluster Size Test Group 1 Group 2 Group 3 Group 4 Average Variance Average Variance Average Variance Average Variance K=2 smoking 43.40769 1246.882 20 1250.73 - - - - gender (m=0 f=1) 0.584615 0.246635 0.526316 0.263158 - - - - age 60.5 119.8524 65 124.2515 - - - - max tumor % 60 800 73.42105 169.5906 - - - - K=3 smoking 43.25 1350.064 53.7 1277.596 42.90625 1077.776 - - gender(m=0 f=1) 0.5 0.26087 0.555556 0.261438 0.619048 0.24158 - - age 62.59524 90.53949 63 104.5882 62.625 194.2446 - - max tumor % 72.02381 306.1702 73.05556 176.8791 64.58333 315.0362 - - K=4 smoking 43.84375 1342.491 41.42857 1190.11 52.44706 1327.423 47.35 609.1125 gender (m=0 f=1) 0.604167 0.244238 0.5 0.269231 0.588235 0.257353 0.4 0.3 age 61.85417 119.9145 63.21429 88.33516 63.64706 103.1176 66 354.5 max_tumor % 69.6875 347.2407 65 276.9231 73.82353 176.6544 76 80

---

Questions or Comments?

Email Me! Emily K. Mower