Thursday, October 20, 2016

Data Analysis Tools - Pearson Correlation


Pearson Correlation

This week I ran a Pearson Correlation for the depth and diameter variables. Running a scatter plot to determine whether there was a positive relationship between crater  diameter and depth was not as clear as I'd like when done for all ejecta types. I limited the variables to the "MLER" top layer ejecta type and produced the following plot.



Definitely positive and fairly linear. But is it significant? Here's the Pearson Correlation for the plot above.




As you can see, the p value is <.0001 and the r value is 0.78221 so I can safely say that for this ejecta type, there is a moderately strong positive linear relationship between crater diameter and density.

Squaring the coefficient gives a value of .611852 and so it can be assumed that 61 % of the variability of crater depth can be predicted.

Here is the SAS code used.

/* opens library containing data */
LIBNAME mydata "/courses/d1406ae5ba27fe300 " access=readonly;

/*pulls specific dataset from library */
data new;
            set mydata.marscrater_pds;

            /* Group ejecta variable by top layer */
            if index(MORPHOLOGY_EJECTA_1, '/')> 0 then
                        Ejecta_1=substr(MORPHOLOGY_EJECTA_1, 1, index(MORPHOLOGY_EJECTA_1, '/')-1);
            else
                        Ejecta_1=MORPHOLOGY_EJECTA_1;
            Ejecta=compress(substr(upcase(Ejecta_1), 1, 4));

            /*Recode crater depth to meters from kilometers*/
            depth_m=depth_rimfloor_topog*1000;

            /*Limit to top layer ejecta type "MLER"*/
            if ejecta='MLER';

            /*Limit to craters 50 km or less in diameter.*/
            if diam_circle_image LE 50;

proc sort ;
            by crater_id;

            /*Produce scatter plot to asses linearity.*/
PROC GPLOT ;
            PLOT depth_m*diam_circle_image;

            /* Generate correlation coefficient. */
PROC CORR ;
            VAR depth_m diam_circle_image;

RUN;

No comments: