Thank you Yury! Super intuitive and straightforward for beginners! It’s so hard to find an introductory video that doesn’t start with functions and notations. And very inspiring, providing me with lots of new ideas on my own research! One piece of personal advice is that I prefer not to be distracted by background music when learning some serious stuff.
Glad you enjoyed it, Grace! I myself came from non-math background and didn't find any good explanations. So, I first needed to explain it to myself and then sought it might be useful for others :) And thanks for the feedback! You are right (and not the first who said that :), the music there is loud and with the combination with my non-native-english is horrible. It was my first video with music though. Since then, I hope, it got better. I improved the sound recording and turned down the volume. If you can check one of my last videos, i.e. on paired wilcoxon test, or correlation, or chi-squared test, and give me another feedback, whether it improved or not, I would highly appreciate that! Have a nice day, Grace! Cheers
Great tutorial! Please continue your work, it's awesome. For me as a not native Englishman the music in background brought some problems. Besides, please check your mic to decrease the reverberation. The quality of your text and visualization are great, thank you.
Thank you very much Vladimir!!! It motivates! An extra thanks for the constructive feedback! I'll do my best to increase the sound quality and to upload good subtitles. Kind regards!
Thanks for the great explanation of the not-to-simple theme! I am trying to understand how to apply this approach to HR analysis in the company. Recently, I found the article, and I was confused about subsetting origin data set through accounting "current time of event minus 12 months". I will be very grateful to hear your thoughts!
Glad it was helpful! I have no idea what that paper meant with -12 months :) and why they did that. Hmm, not every scientific paper is perfect and to be trusted. Thus, either we both don't know what they talk about, or they did not explained it right, or they just were lucky with reviewers.
Do you plan on covering parametric extrapolations of survival curves? Do you plan on explaining the logic behind applying hazard ratios to survival curves to, for example, compare survival of patients on different cancer drugs?
Hey Marcus, the very next video is on parametric (exponential) survival models. It might not be too deep, but it's a beginning I hope. Yes, I plan to cover more, including hazard ratios and cox models. It would take time however, because of the day job and because I wanna cover some basics, which I would have loved to have as I started data science, like p-values etc. So, stay tuned and happy learning till then. Cheers
Is it ok to use KM without censoring? like if I know the reason of dissapearance (e.g. in a experiment one organism eat the other), therefore is dead. Or should I use other type of analysis?
HI, thanks for a videos! I have a one question. What should we do with patient who was lost in follow-up withoun any information after discharge? So in the column "time" I litterally dont know what to type. All I know he was alive at the start of the my study, no contact at all. Should I type 1 day or exclude the patient? Becasue I think that by typing 0 in "event" and 1 in "time" variable I just overestimate the survival.
Hi, you are welcome. Since the patient was alive, 0 in event is correct. And since it was alive the last time you have seen it, even only day 1, it's ok to write the day 1. Such patient is called - censored. It was alive the last time we saw it, even if it is just one (first) day. However, if you have a lot of such patients in your data, they will not produce meaningful inference, so you might consider to exclude them.
Great tutorial, it's very useful and amazing, thanks a lot for that, excuse me, I try to run the commands on R but i always get : Error in Surv(time = d$time, event = d$status) : object 'd' not found, could you please help me?
Hey Amr, thanks! Sure, "d" is your dataset. both, time and status, your columns. So, load your data into R and, either call it "d" or change "d" to the name of your dataset. Hope that helps! Cheers
Great work. You explained it with a fun example and included R analyses - icing on the cake !
Glad you liked it! Thanks you for feedback and for watching!
Thank you Yury! Super intuitive and straightforward for beginners! It’s so hard to find an introductory video that doesn’t start with functions and notations. And very inspiring, providing me with lots of new ideas on my own research!
One piece of personal advice is that I prefer not to be distracted by background music when learning some serious stuff.
Glad you enjoyed it, Grace! I myself came from non-math background and didn't find any good explanations. So, I first needed to explain it to myself and then sought it might be useful for others :) And thanks for the feedback! You are right (and not the first who said that :), the music there is loud and with the combination with my non-native-english is horrible. It was my first video with music though. Since then, I hope, it got better. I improved the sound recording and turned down the volume. If you can check one of my last videos, i.e. on paired wilcoxon test, or correlation, or chi-squared test, and give me another feedback, whether it improved or not, I would highly appreciate that! Have a nice day, Grace! Cheers
I would love to learn all statistics from you. Your teaching is just amazing. I never seen anyone to gone that much depth.Have to appreciate.
So nice of you! Thousand thanks!
Very nice and clear presentation of survival analysis (or failure analysis). Thanks!
Glad you liked it! Sorry for a pure sound quality, this was one of my first videos. They are better nowadays I hope.
Great tutorial! Please continue your work, it's awesome. For me as a not native Englishman the music in background brought some problems. Besides, please check your mic to decrease the reverberation. The quality of your text and visualization are great, thank you.
Thank you very much Vladimir!!! It motivates! An extra thanks for the constructive feedback! I'll do my best to increase the sound quality and to upload good subtitles. Kind regards!
Thank you sir. This is very helpful.
Thanks for your feedback!
Thanks for the great explanation of the not-to-simple theme! I am trying to understand how to apply this approach to HR analysis in the company. Recently, I found the article, and I was confused about subsetting origin data set through accounting "current time of event minus 12 months". I will be very grateful to hear your thoughts!
Glad it was helpful! I have no idea what that paper meant with -12 months :) and why they did that. Hmm, not every scientific paper is perfect and to be trusted. Thus, either we both don't know what they talk about, or they did not explained it right, or they just were lucky with reviewers.
Thank you so much. Its a great tut.
Glad it was helpful!
Do you plan on covering parametric extrapolations of survival curves? Do you plan on explaining the logic behind applying hazard ratios to survival curves to, for example, compare survival of patients on different cancer drugs?
Hey Marcus, the very next video is on parametric (exponential) survival models. It might not be too deep, but it's a beginning I hope. Yes, I plan to cover more, including hazard ratios and cox models. It would take time however, because of the day job and because I wanna cover some basics, which I would have loved to have as I started data science, like p-values etc. So, stay tuned and happy learning till then. Cheers
Is it ok to use KM without censoring? like if I know the reason of dissapearance (e.g. in a experiment one organism eat the other), therefore is dead. Or should I use other type of analysis?
it sounds to me still as censoring... the main thing is - the animal did not die from the reason you are studiying... but from the different reason.
HI, thanks for a videos! I have a one question. What should we do with patient who was lost in follow-up withoun any information after discharge? So in the column "time" I litterally dont know what to type. All I know he was alive at the start of the my study, no contact at all. Should I type 1 day or exclude the patient? Becasue I think that by typing 0 in "event" and 1 in "time" variable I just overestimate the survival.
Hi, you are welcome. Since the patient was alive, 0 in event is correct. And since it was alive the last time you have seen it, even only day 1, it's ok to write the day 1. Such patient is called - censored. It was alive the last time we saw it, even if it is just one (first) day. However, if you have a lot of such patients in your data, they will not produce meaningful inference, so you might consider to exclude them.
Time (days) is a categorical variable right?
Hey Martin, time is always numeric.
Great tutorial, it's very useful and amazing, thanks a lot for that, excuse me, I try to run the commands on R but i always get : Error in Surv(time = d$time, event = d$status) : object 'd' not found, could you please help me?
Hey Amr, thanks! Sure, "d" is your dataset. both, time and status, your columns. So, load your data into R and, either call it "d" or change "d" to the name of your dataset. Hope that helps! Cheers
@@yuzaR-Data-Science Thank you so much :)
@@amrsalaheldin3814 Very welcome!