too much data heterogeneity

1. difficulty level of getting into top 5 vs. the rest of top20-30 is very different. but the data of addimision to top 5 is thin even for those top high schools. so depending on your target, need to separat the datapoints for different purpose

2. addmission to top 5 is too thin, some legacy addimission will make it hard to draw meaningful conclusion. 

agreed with others:

 關鍵是看你自己孩子的性格和實力，承壓能力強不強。你說的這些學校對你孩子的升學概率沒什麽大的差異