Understanding model interpretability in R with ggplot2 and mikropml (CC134)

[@Riffomonas]

Would you be interested in seeing me do an interactive live stream episode? It would likely run longer than normal but would give you all a chance to make comments and ask questions about using the concepts I've been covering. Please "like" this comment to tell me what you think! If I get enough likes, I'll be sure it happens. It will likely happen around noon Eastern US time

[@che-weilu334]

I am a new user of mothur to analyze the environmental consortia of my Ph.D. projects. I learned the tidyverse and ggplot skill on the Datacamp by myself. Unfortunately, it took me around a whole year until I saw this YT channel. Learned a lot from every episode. Glad to see the live stream episode.

[@heshamamin2765]

Thanks a lot for package and the videos. I just have question. When your outcome contain two charterer (health and srn) . How to do you know that the weights from the models either positive and negative reflecting on the srn not health. let say that my outcome have two charter (normal and abnormal) with equal number of observation when I make plot like yours with median of weights. How I can be sure that this features related to abnormal observation either positively or negatively.

[@Riffomonas]

THanks for watching! It's great you're trying to run this with your own data. I'm pretty sure my model was detecting SRN because that column was a logical. So if the value was TRUE, then that would mean it was originally srn and if it were FALSE, then that would mean it was healthy/normal. You need to create this column as a logical.

[@johansebastian88]

Dear Patrick, Thank you for this great series of videos. I was able to use my own data however I am still confuse about something. How do you know to which of your conditions the features are positive or negative associated. is because the alphabetical order? Cheers!!

[@Riffomonas]

Yeah you can code the variable as 0 or 1 or put it in alphabetical order. You can also always subtract the value from 1 if they're all in the wrong direction.