Comments - Simple Trick to Normalize Correlations, R-squared, and so on - Data Science Central2021-06-15T21:42:23Zhttps://www.datasciencecentral.com/profiles/comment/feed?attachedTo=6448529%3ABlogPost%3A833355&xn_auth=noThanks, this is very practica…tag:www.datasciencecentral.com,2019-07-02:6448529:Comment:8510262019-07-02T11:10:11.007ZJonathan Nathanhttps://www.datasciencecentral.com/profile/JonathanNathan
<p>Thanks, this is very practical and useful. I hope I got it right. </p>
<p></p>
<p>I allow myself to suggest one didactic pointer:</p>
<p>The final result is not stated clearly enough (for me). I would write (and I hope I got it right!) a concluding line such as:</p>
<p>So now, instead of making an invalid comparison between the correlation of different number of samples (0.92 for 20 samples compared to 0.3 of 10 samples), we can make a valid comparison. We compare the mean correlation of the…</p>
<p>Thanks, this is very practical and useful. I hope I got it right. </p>
<p></p>
<p>I allow myself to suggest one didactic pointer:</p>
<p>The final result is not stated clearly enough (for me). I would write (and I hope I got it right!) a concluding line such as:</p>
<p>So now, instead of making an invalid comparison between the correlation of different number of samples (0.92 for 20 samples compared to 0.3 of 10 samples), we can make a valid comparison. We compare the mean correlation of the 10 sample experiments on the 20 sample dataset (0.76) to the correlation of the 10 sample dataset (0.3).</p>
<p>Thanks again for this and more of the useful methods you publish!</p>