How to check for correlations among rasters and keep only those that are not highly correlated

[@KnorpelDelux]

Great video! It really helped me with understanding the concept

[@gorderos]

Great video! thank you.

[@JoannaLyall-vm7bl]

Hi Josh, loved the video and clear explanations so I had great expectations when I ran my rasters for correlations. Initially I tried to run them all together which took far too long and resulted in >50 warning messages e.g. "1: In cor(df[, 1], df[, 2], method = method) : the standard deviation is zero". I have tried running a subset of 10 rasters but same result. The area they cover is approx 6.81 million hectares. They work in MaxEnt but I wanted to check for correlations before running again for clearer results. How can I get the correlation check to work?

[@DrJBanta]

Hello, thank you for the positive feedback.Some things to try: (1) Did you set the working directory to the folder that contains your rasters? Remember that the rasters cannot be in a subfolder. They have to be directly in the working directory. (2) Did the rasters import into R successfully? Type summary(bio1) and hit enter. If you called your object something other than bio1, type in the name of your object. (3) You may have run out of RAM memory. In that case, try the standalone version of ENMTools. I believe it is already installed on my Virtualbox virtual machine, but if not, here are instructions: kzbin.info/www/bejne/g5uWm6JnZceAZ80 . To run the correlations check using the correlations check using the stand-alone software, see these instructions: drive.google.com/file/d/1q3bT0eOsXb2Ucgsw0RJdymFsRTrWZbKL/view?usp=sharing . Best, Josh

[@JoannaLyall-vm7bl]

@@DrJBanta Thank you for your reply. I have run the first set of rasters again, double checking I have the working directory set. It took a long time but eventually finished again with the warnings. As you recommended, I checked the imports were successful and for the first one I had these results: bioclim_tasmania_SILO_1980.1999_p01 Min. 4.007800e+00 1st Qu. 9.271800e+00 Median 1.074640e+01 3rd Qu. 1.176283e+01 Max. 1.440070e+01 NA's 1.452405e+07 Warning message: In .local(object, ...) : summary is an estimate based on a sample of 1e+05 cells (0.47% of all cells) which I believe indicates the file was successfully imported. Would the NA's be causing the issue?

[@haticeozkan1331]

Dear Josh Banta, I follow your videos. Your videos are very helpful. Could you make a video explaining how we can make graphs like "von bertalanffy growth equation" and "annual growth rate" (spss, excel)?

[@bangsawinata]

Hello Prof. Previously I am very grateful because you have provided various video tutorials for using MaxEnt. I would like to ask about correlation analysis, in my case, I use raster data with continuous and categorical variables. After I read some papers he said categorical data should use another analysis (chi-square). Do I have to keep doing correlation analysis or not before using MaxEnt?

[@DrJBanta]

Hello, I do not think chi-squared analysis would work in this case for your categorical data. I typically leave out categorical data from the correlation analysis, and then just include all of the categorical data in the modeling effort.

[@bangsawinata]

@@DrJBanta Thanks for the answer

[@ashamerin]

nextgenerationsequencinghq?