The correlation plot looks even better with one small tweak. Add a diverging setting to the conditional formatting and set it to a white color. This way, non-correlating items will be white instead of purple. The PBIX download already has this change.
I was thinking the same thing while watching the video. Nice job on this. It's amazing what you can do with some patience and understanding of how to make things work the way you want them to.
I'm trying to get this to work I followed your video but when I add the correlation coefficient it's only displaying a number where x = y, is there a way to get it to display the calculation for everything else as well?
@@khushaalchaudhary6955 I got it to work by looking at the relationships between the tables/attributes tables. If you download the example file and ensure your relationships between the attributes tables/correlation tables are the same it should work!
Hey man - I'm trying to make a tooltip on top of this. I've made a tooltip page in order to achieve it, and managed to get a header that (when hovering on a specific cell) displays the "[x variable] v [y variable]".. But the scatter plot within the tooltip itself ends up empty (unless it is the cell where x is the same as y). Any ideas on what my measures for X and Y should be on the scatter plot that simply needs to display the filtered data? Thanks so much.. My x measure so far is this btw (but it doesn't work!) Xall = VAR CurrentX = SELECTEDVALUE(Name2[Column1]) RETURN CALCULATETABLE( SELECTCOLUMNS( FILTER( 'Peers', 'Peers'[ISIN] = CurrentX ), "pChg", 'Peers'[pChg] ) )
Great video! If anyone is following this, I would highly suggest removing outliers if you can. The Pearson r correlation is very sensitive to outliers and you might drown out your results if you don't remove them. A good way to remove outliers is to calculate your average and standard deviation for each attribute and then remove anything that isn't within (above or below) 3 standard deviations of that average. This can be quite tedious with larger datasets but it's worth it for finding those relationships between your variables.
Hi, Thanks for the video. Please, may you share the source file for the dataset used in this video, because the link you share on youtube description does not includes any excel files.
Great video. Thanks for sharing. However, my correlation coefficient return abnormal number, same attribute in X and Y do not return 1 but 0.xx instead. Has anyone face similar situation?
This could not have been better timed. I will be using it tomorrow to create a simple matrix like this. After that, if the results are as expected, I will be venturing into Python land to make a differen depiction where "r" will be controlling the size of the box. It's very hard to perceive color gradient to evaluate the correlation, but size makes a better "right in your face" representation.
Great technique and knowledge.Thank you. Btw,Why should we use MAX function in a CALCULATE function. I succeeded making the matrix but still have a question.Does anyone have an idea.
Hello, thank you so much to present this solution. Please, assuming that this matrix are with the same results spitted into two triangles, could you tell me how can we display only the one of those? Best, Rafael
thanks for the video.... i am actually needing some solution similar to this.... the video looks promising, but its really not working for me. I am being left with "1.00" acros the whole matrix
Hi i have implemented similiar code, somehow i end up getting only one's in the diagonal but no values anywhere else, not even zeros, did some one get , where exactly i might be making a mistake?? Thanks.
looks great! I added a slicer for a hidden column that has a date_month but it doesn't seem to be filtering the data, do I have to adjust the measures to take the filter?
Hello sir, very helpful video thanks for that. Also i have a question, im looking for this matrix visual that you made but i need to make row and column makes effect from separate date filters i mean, i need 1 date filter effect rows and 1 date filter effects columns, how can i do it ?
I'm somewhat new to this so my question is...is there any reason why this couldn't all be done in one measure using more variables instead of each measure creating the virtual table and iterating over it?
I just re-created a matrix following the steps and it didn't work because of the 'behind-the-scenes' linkages. If the matrix feels that it is presenting two sets of related data (related using joins) it only presents the diagonal axis. I had to delete the table associations for the matrix to work. I think, therefore, that using the same set for both axis would give the same result. It's an incontinence of PowerBI but one we can get around fortunately.
@@TemiAnData the main power bi screen had 3 icons in the left hand side. Report view, Data view and Table relationships. Go into table relationships and you'll notice your generated category tables are linked back to the table they were generated from. Delete the two associations. Done!
Very helpful, thank you! (I haven't used virtual tables much and am still getting a grip on passing values in and out of variables so this was an awesome example to play with)
Fantastic video! I tried to follow its steps, but I couldn't because it didn't show all the calculations of the "Cars Pivot" (Index, Attribute, Sum Value). Could you send the formulas for each "Cars Pivot"? Thank you very much!
Reviewing the video, I understood that a virtual table was created with the attribute and index, but I couldn't understand the measure of sum "value". Could you inform the calculation?
Unbelievable that you have to roll this yourself in Power BI! This is a one-liner in Python's Pandas library. df.corr().style.background_gradient(cmap='RdBu', vmin=-1, vmax=1)
Yes,you have to work on the data modelling part..just download the corelation calculation file & follow the same..you will get the desired result..let me know if you able to solve it or not 😊😊😊
@@BIElite Pleeeeease if you find a way let us know, this is the best video I've seen about correlation matrix in Power BI, but I could't use it since all my data is in direct query mode, I'm so sad.
