Dear Mr.Gyawali, thank you very much for sharing this comprehensive video. I have one question regarding how the AUC curve is made. Initially, you have mentioned you are going to use the training and testing data to make the AUC curve. However, I haven't seen you using the testing data while making the AUC curve. You have only used the training data set. Would you please explain this? Thank you very much in advance!
Yes you are absolutely right about this ...you should use testing data instead of training data..training sample is for map making process but to check its accuracy using auc graph we should use testing sample.
The video is helpful. For calculation of AUC, i classified the susceptibility map into 100 classes and reclassified it, for tabulate area i added the reclassified map and training data with the field having 1. However, i get 52 values not 100 values, why is that? how can i calculate AUC with these values or something wrong with the result? Of course in your case also i saw values from 18 to 99, and then in other excel 1-100. How do you do that? Am waiting for your answer. Thank you in advance.
Thank you so much for this video. I have a small question and hope you will answer it. Can I use the Max-Min method for calculating factor weight (PR value) when I use the other bivariate statistical methods? For example, I used Landslide susceptibility analysis method (LSA) and Statistical index method (SI) for my analysis but when I calculated factor weights using Max-Min method, the results seem do not reasonable.
Hi pls answer this thanks very much. Can I use the same method to obtain the Prediction Frequency (PF) and apply it to models such as Weights of Evidence (WoE)?
Thanks for your videos sir. I have a question, regarding the separation of landslides into 30% for testing and 70% for training, to make the statistical models should I use 70% of the training section or should I use 100% of the landslides in polygons?
landslide polygon is taken from field survey or Google earth and later we divided in to 70/30 by using random sampling tool in gis...so 70% landslide polygon is training and 30% are taken as testing
@@Geoinfonepal for example I make final susceptibility map with training data.. And get 75 % accuracy... So for testing... I wil used testing data again like to put in final susceptibility map.. And get some results.. Like 70 %.. Is it..
Thanks for such a nice video, after drawing up the landslide inventory in the study area, and using google earth to make the landslides, and then how I can divide the results in two: for example 70 % for training and 30% for taking the ArcGis test? Tank you Mr
Really appreciate your work..Thank you Mr.Gyawali.Can you make another video on weight of evidence model for susceptibility mapping.It will be great if you speak while making video,it will be easy to understand easily.
Plz let me know... Why didn't put new values in reclassify raster of LULC... Bcoz at 19:00...in final raster calculation it seem you put raster value without this🙂. I m waiting for your answer 👍
i gave the new generated fr weight for reclassify after that in raster calculation i used this reclassified data to generate final map using raster calculator
thank you for an amazing video. Mr. Gyawali my RF values are going high(>7) for the buffer (apply different distance) of road and stream. why is it? kindly give some suggestions to overcome this fault. thank you
it always depends upon area and landslide pixel same as area depends buffer zone.you should take buffer area according to your study area and with the help of literature review...if your obtained value is wrong then iterate again.
Hello. Thancks for the video. Could you please tell me which article have you used to obtain the methodology to calculate the Prediction Rate (PR)? Thank you in advance.
Do you have any reference on LOGISTIC REGRESSION and SUPPORT VECTOR MACHINE methods of Landslide susceptibility mapping? how to apply these models? how to use these models for LSM?
after conversion of the training and testing data into raster how to make it of one division as n TOC coz in my case it shows the total samples of training and testing
Dear Sir, I would love to know, In Part-I, you used a different formula for FR whereas, that Column is totally absent in Part-II, rather we see a column with the same name (FR) that has been using a different formula (Percentage/Percentage). It has made me a bit confused. Should I ignore the first video and focus only on the Excel of Part-II (this video)?
Hello. Thank you for the very informative tutorial. For the 11th column (RF INT) since most of my RF % values are more than .5, eg 60.80 but RF gives 60 instead of 61 thereby the sum is not 100. Is there a problem if I round them off manually achieve the sum of 100%. Thank you.
Dear Pradeep Ji, Please make a video on AUC curve with details. It will be really helpful to all!! Thank you in advance. The steps in this video are not clear.
Hi, Sir! Thank you for your explanations! But, I actually have a questions regarding the AUC calculations.. in the excel, your OBJECT_1 coloumn really has a high value. is it just using tabulate the classified 100 classed LSI map with training data? because when I did that, I really just get the same total value with my sum landslide pixel of each parameter. its the exact same number🥲 is that actually okay? because I think I got low AUC valie because of those number😭 ah, and I already put the same cell size for each parameter which is 10m.. Maybe you can help me, sir? Thank you, sir! Best Regards, Putri
At 25:47 minutes, you are tabulating the datasets for AUC. While you are tabulating the area using "Reclass_final.tif"( under Input raster zone data) and training_ras.tif (under Input raster or feature class data). Should we be using "Testing.tif" instead of "Training.tif"? I am not clear on the segment of the video. Glad if you could answer me.
I am sorry Mr.Tenzin, I am too late for your reply. I am happy for your question, For tabulating the datasets for AUC it must required testing.tif instead of traing.tif dataset so use testing.tif...there is a mistake. Thank you
Sir, I have a question what is the correct formula for FR value 1. landslide pixel (each class) / Class pixel (each class) or 2. % of landslide pixel (each class) / % of class pixel (each class). I am confused in 1st part to obtained FR value you used first process which mentioned above and 2nd part you use second process. please reply.
1, Frequency Ratio = % of landslide pixel (each class) / % of class pixel (each class). OR 2. Frequency Ratio = (a / (a + b)) / (c / (c + d)) In this formula: 'a' represents the number of events (e.g., occurrences of a specific phenomenon) that occurred within a specific condition or area of interest. 'b' represents the number of non-events (e.g., absence of the phenomenon) within the same condition or area of interest. 'c' represents the number of events that occurred outside the condition or area of interest. 'd' represents the number of non-events that occurred outside the condition or area of interest. in formula 1 (a+b=1 and c+d=1 )
Dear, GeoinfoNepal. I'am so confused right now of this part 2 video. I feel skip on the part that I don't know how the datas come up on the table. Can you make an video of this how we get all of this data that input in your excel. It doesn't match the datas on the previous excel which is the part 1 of this.
yeah there might be difference in data in 1st and 2nd part of this video...main thing is that i want to show the concept of it...so exactly in which part, you confused to get all the data in excel file
excuse me sir i would like to know how you made that final frequency table. also how do you calculate the class pixes and the % class pixels because i feel you skipped all those steps when making this video because in the last video you did not have this table.
2 videos nikali hai aapne aur akhri mai AUC vaale mai kya kar diya sir ji.... kuch samjh hi nahi aaraha.. please explain how you got those values after tabulate area process of LSM and traininig data.
The video has really been helpful. Thank you Mr. Gyawali
welcome Ms. Joshi 😂
Thank you for the valuable video ☺
Dear Mr.Gyawali, thank you very much for sharing this comprehensive video. I have one question regarding how the AUC curve is made. Initially, you have mentioned you are going to use the training and testing data to make the AUC curve. However, I haven't seen you using the testing data while making the AUC curve. You have only used the training data set. Would you please explain this?
Thank you very much in advance!
Dear Mr. Gyawali, Would you please answer my question? I don't understand why you were not using the testing data when you make the AUC curve.
Yes you are absolutely right about this ...you should use testing data instead of training data..training sample is for map making process but to check its accuracy using auc graph we should use testing sample.
@@Geoinfonepal Thank you very much!
@@selamawitdamtew3583 Does this mean that we use training datasets for tabulating area rather than using the original landslide data?
The video is helpful. For calculation of AUC, i classified the susceptibility map into 100 classes and reclassified it, for tabulate area i added the reclassified map and training data with the field having 1. However, i get 52 values not 100 values, why is that? how can i calculate AUC with these values or something wrong with the result? Of course in your case also i saw values from 18 to 99, and then in other excel 1-100. How do you do that? Am waiting for your answer. Thank you in advance.
yes please answer this, im confused as well. thanks in advance
Can you please upload video on ROC validation for information value method..
Thank you!
Thank you so much for this video. I have a small question and hope you will answer it. Can I use the Max-Min method for calculating factor weight (PR value) when I use the other bivariate statistical methods? For example, I used Landslide susceptibility analysis method (LSA) and Statistical index method (SI) for my analysis but when I calculated factor weights using Max-Min method, the results seem do not reasonable.
thank you so mach for this video.but can you tell me how to get RF(INT) value? and can PR value greater than 100?
Hi pls answer this thanks very much. Can I use the same method to obtain the Prediction Frequency (PF) and apply it to models such as Weights of Evidence (WoE)?
thanku brother ..😊
Thanks for your videos sir. I have a question, regarding the separation of landslides into 30% for testing and 70% for training, to make the statistical models should I use 70% of the training section or should I use 100% of the landslides in polygons?
Thanks for such a nice video..Please tell me how to make training and testing on what basis?is it observed data?
landslide polygon is taken from field survey or Google earth and later we divided in to 70/30 by using random sampling tool in gis...so 70% landslide polygon is training and 30% are taken as testing
@@Geoinfonepal Thanks for your prompt reply...Can you make video, how to make training and testing data? stay safe sir
is it possible to calculate susceptibility of area where there is no landslide enregistred in the past tahnkyou for answering sir
Hi, used here only training data...and testing data?
Testing data is used for accuracy assessment and to generate AUC/ROC curve.
@@Geoinfonepal for example I make final susceptibility map with training data.. And get 75 % accuracy... So for testing... I wil used testing data again like to put in final susceptibility map.. And get some results.. Like 70 %.. Is it..
Thanks for such a nice video, after drawing up the landslide inventory in the study area, and using google earth to make the landslides, and then how I can divide the results in two: for example 70 % for training and 30% for taking the ArcGis test? Tank you Mr
my next tutorial is there for your this question. Please check there.
@@Geoinfonepal Thank you
Really appreciate your work..Thank you Mr.Gyawali.Can you make another video on weight of evidence model for susceptibility mapping.It will be great if you speak while making video,it will be easy to understand easily.
okay Mr.thapa i will do ...thank you for your comment
@@Geoinfonepal Helo Mr.Gyawali, when will you making another video on weight of evidence model for susceptibility mapping.wating for your next video.
Hello Mr.Thapa...due to my busy schedule i cant make WoE till now but i will do it very soon...keep in touch...Thank you😊😊
@@Geoinfonepalok,hope you will upload soon.thank you
Hi can you make a short tutorial on Logistic Regression Model? Or how we can use SPSS to run these kind of analysis?
If anybody know the Logistic Regression Method, Kindly upload.
Plz let me know... Why didn't put new values in reclassify raster of LULC... Bcoz at 19:00...in final raster calculation it seem you put raster value without this🙂. I m waiting for your answer 👍
i gave the new generated fr weight for reclassify after that in raster calculation i used this reclassified data to generate final map using raster calculator
@@Geoinfonepal thank you for your reply brother
thank you for an amazing video. Mr. Gyawali my RF values are going high(>7) for the buffer (apply different distance) of road and stream. why is it? kindly give some suggestions to overcome this fault. thank you
it always depends upon area and landslide pixel same as area depends buffer zone.you should take buffer area according to your study area and with the help of literature review...if your obtained value is wrong then iterate again.
Very nice.please tell me how to make training raster.tif
Hello. Thancks for the video. Could you please tell me which article have you used to obtain the methodology to calculate the Prediction Rate (PR)? Thank you in advance.
iopscience.iop.org/article/10.1088/1755-1315/280/1/012005/pdf
How did you classify plan curvature into concave, flat and convex. Is 0.05 a fix number? Thank you!
Please make short video how to create training raster tif
Do you have any reference on LOGISTIC REGRESSION and SUPPORT VECTOR MACHINE methods of Landslide susceptibility mapping? how to apply these models? how to use these models for LSM?
Is this sensitivity vs 1-specificity? Opps!
after conversion of the training and testing data into raster how to make it of one division as n TOC coz in my case it shows the total samples of training and testing
you need to divide it in to 2 parameters of your total landslide pixel using statistical sample manager
the whole video was very informative but I did not understand how you did AUC? can you please make a detailed video on AUC?
Please visit my next tutorial for AUC curve generation
@@Geoinfonepal please sir make details video in this.....also provide me Excel calculation sheet if possible
bro.. can u tell me how u calculate excel values??? i mean that FR, RF , RF(non%)...
can i use RF% values for calculating LS instead of INT RF
Dear Sir,
I would love to know,
In Part-I, you used a different formula for FR whereas, that Column is totally absent in Part-II, rather we see a column with the same name (FR) that has been using a different formula (Percentage/Percentage). It has made me a bit confused. Should I ignore the first video and focus only on the Excel of Part-II (this video)?
can you please upload video on fire hotspot mapping and risk assessment with basic details? :) thank you in advance.
Hello. Thank you for the very informative tutorial. For the 11th column (RF INT) since most of my RF % values are more than .5, eg 60.80 but RF gives 60 instead of 61 thereby the sum is not 100. Is there a problem if I round them off manually achieve the sum of 100%. Thank you.
Okay thank you for your commemt
Nice, I like it, can you let me how to get landslide data?
LANDSLIDE polygons are digitized from satellite image/aerial photo or google earth
@@Geoinfonepal Thanks
sir, my PR values are exceeding 15, is that okay?
Dear Pradeep Ji, Please make a video on AUC curve with details. It will be really helpful to all!! Thank you in advance. The steps in this video are not clear.
Dear sir, Here is the steps for preparing AUC/ROC curve: ua-cam.com/video/xU0ExUj0PZU/v-deo.html&ab_channel=GeoinfoNepal
Hi, Sir!
Thank you for your explanations! But, I actually have a questions regarding the AUC calculations..
in the excel, your OBJECT_1 coloumn really has a high value. is it just using tabulate the classified 100 classed LSI map with training data? because when I did that, I really just get the same total value with my sum landslide pixel of each parameter. its the exact same number🥲 is that actually okay? because I think I got low AUC valie because of those number😭
ah, and I already put the same cell size for each parameter which is 10m..
Maybe you can help me, sir? Thank you, sir!
Best Regards,
Putri
Sir, without WOE method part 3 , Woe method is incomplete
where you came the value 0.169?
At 25:47 minutes, you are tabulating the datasets for AUC. While you are tabulating the area using "Reclass_final.tif"( under Input raster zone data) and training_ras.tif (under Input raster or feature class data). Should we be using "Testing.tif" instead of "Training.tif"? I am not clear on the segment of the video. Glad if you could answer me.
I am sorry Mr.Tenzin, I am too late for your reply. I am happy for your question, For tabulating the datasets for AUC it must required testing.tif instead of traing.tif dataset so use testing.tif...there is a mistake. Thank you
Where does the training raster come from?
From field data or from secondary source.
Sir, I have a question what is the correct formula for FR value
1. landslide pixel (each class) / Class pixel (each class) or
2. % of landslide pixel (each class) / % of class pixel (each class).
I am confused in 1st part to obtained FR value you used first process which mentioned above and 2nd part you use second process.
please reply.
1, Frequency Ratio = % of landslide pixel (each class) / % of class pixel (each class). OR
2. Frequency Ratio = (a / (a + b)) / (c / (c + d))
In this formula:
'a' represents the number of events (e.g., occurrences of a specific phenomenon) that occurred within a specific condition or area of interest.
'b' represents the number of non-events (e.g., absence of the phenomenon) within the same condition or area of interest.
'c' represents the number of events that occurred outside the condition or area of interest.
'd' represents the number of non-events that occurred outside the condition or area of interest.
in formula 1 (a+b=1 and c+d=1 )
Dear, GeoinfoNepal.
I'am so confused right now of this part 2 video. I feel skip on the part that I don't know how the datas come up on the table. Can you make an video of this how we get all of this data that input in your excel. It doesn't match the datas on the previous excel which is the part 1 of this.
yeah there might be difference in data in 1st and 2nd part of this video...main thing is that i want to show the concept of it...so exactly in which part, you confused to get all the data in excel file
Hello sir, I still don't get how you made the final frequency ratio table.
i will share my table if you provide your mail address.
@@Geoinfonepal That would be great help. Please email it at mandalswati113@ gmail dot com. Thanks 😇
How you make testing and training data?
using statistical sample manager
Sir please tell me the paper for AUC used here .
www.researchgate.net/publication/356422512_Landslide_Susceptibility_Assessment_Using_Bivariate_Statistical_Methods_A_Case_Study_of_Gulmi_District_western_Nepal
I don't understand how do you get the AUC excel sheet's value and object_1?? Plz kindly clarify my doubt by replying me..Thank you🙏🏻
ua-cam.com/video/2XAhjJInDiI/v-deo.html... using point data for auc curve generation
@@Geoinfonepal Thank you so much
excuse me sir i would like to know how you made that final frequency table. also how do you calculate the class pixes and the % class pixels because i feel you skipped all those steps when making this video because in the last video you did not have this table.
ua-cam.com/video/X5ho-1Iqn6E/v-deo.html&ab_channel=PradeepGyawali
2 videos nikali hai aapne aur akhri mai AUC vaale mai kya kar diya sir ji.... kuch samjh hi nahi aaraha.. please explain how you got those values after tabulate area process of LSM
and traininig data.
I don’t understand the part 2. It’s not clear like the first
you can watch my another video(SE method) related to landslide mapping...and i think you can get it
Bad music disturb
WHAT İS PR value
In Frequency ratio (FR), relative frequency (RF) of each class and prediction rate (PR) of each conditioning factor.
PR says the factor weightage
@@Geoinfonepal mu teacher. Can you send me. Your mail adress
@@Geoinfonepal can you send me your mail adress