Supervised ML

[mkholod]

Supervised machine learning Doesn't matter which algorithm Take the params of a single patient, with the DNA age and predict if it has metastasis or not and this is compared to the ability to predict without this param

לא ככ משנה לי איזה אלגוריתם שיקח את הפרמטרים של המטופל יחד עם גיל הדנא ויחזה אם יש או אין גרורות ןזה בהשוואה ליכולת ניבוי בלי הפרמטר הזה

[mkholod]

אפשר יהיה לנסות גם אלגוריתמים אחרים של unsuprevised machine learning כמו k nearest neiboour?

[mkholod]

https://youtu.be/GtgJEVxl7DY?t=816 Values should be normalized to be between 0 and 1

[mkholod]

To try to predict the actual age based on beta

[mkholod]

Part two https://www.youtube.com/watch?v=DkLNb0CXw84

[mkholod]

https://github.com/mkholod/meth/commit/0c981b84ea9db8cca4f1cc2dfa3546fc54dc063f

[mkholod]

• We'll try to predict metastasis instead of an age
• Clustering options -- https://www.geeksforgeeks.org/k-nn-classifier-in-r-programming/
• We can predict age in groups (30-39, 40-49) rather than a descrete number

[mkholod]

Try to predict with supervised the type of the growth (Sporadic / MEN1 / VHL), try KNN

[mkholod]

What's the correlation between accelerations (delta, acc1, acc2 and metastasis) Is there a correlation between deviation between the ability to predict the age and the metastasis

[mkholod]

Apply forrest which takes a lot of subsets for learning