search

vivekaxl / ChinExp

0 stars 0 forks source link

Check the stability of the results #2

Open vivekaxl opened 8 years ago

oxhead commented 8 years ago

Which file/sheet should I check?

vivekaxl commented 8 years ago

I beg your pardon. I was trying to show Ashwin how to assign issues :P

vivekaxl commented 8 years ago

WHAT: Dataset, Median, iqr, dataset 1_tp_read.csv 14.1265 1.2455 32.0 2_tp_write.csv 8.0265 0.6275 32.0 3_tp_read.csv 23.408 1.663 64.0 4_tp_write.csv 13.243 0.246 64.0

Strawman: Dataset, Median, IQR, datapoints 1_tp_read.csv 6.7315 2.10775 70.8 2_tp_write.csv 3.4565 0.63625 71.7 3_tp_read.csv 19.05 3.04275 176.4 4_tp_write.csv 8.2255 1.41325 176.4

vivekaxl commented 8 years ago

@oxhead What do you think?

oxhead commented 8 years ago

Questions:

  1. what is iqr?
  2. dataset --> selected data points
  3. Strawman is someone's approach? Or yours?
vivekaxl commented 8 years ago
  1. IQR: https://en.wikipedia.org/wiki/Interquartile_range: 75percentile - 25percentile
  2. dataset names
  3. strawman (stupid approach): use 30% of the points sampled randomly
oxhead commented 8 years ago
  1. The lower the IQR, the better the stability?
  2. I guess I mean datapoints (your first data sheet might be misspelled?)
  3. From your expertise, WHAT is way better than Strawman?
vivekaxl commented 8 years ago

The lower the IQR, the better the stability?

Correct

I guess I mean datapoints (your first data sheet might be misspelled?)

I don't understand

From your expertise, WHAT is way better than Strawman?

I am not sure. It seems like the space is very simple. I think we need to chec with some more data