Dear all. I'm a new user to this fantastic piece of software. Been using it for a few days, already done some interesting stuff. But right now I am stuck: I am trying to build a co-occurence matrix visualization from a table generated by a "Term co-occourence counter".
Each row of the co-occourence table has a "Term1" and "Term2" columns (of Term type) and one "Document co-occourence" column (integer type) which indicates the number of times Term1 "co-occours" with Term2 in the document.
Question how can I generate a matrix view with all the terms listed in both x and y axis, like this example?
http://dh2016.adho.org/static/data/673/image10.jpg
Thank yiou for your time!
Bruno.
Hi, I found a way to calculate a co-occourence matrix, but using R and the ploting a Heatmap. So you have to "link" Knime with R, so you have to go File/Preferences/R....
Let me know if it works for you.
Best Regards
Thank you mauuuuu5! After installing R, xQuarts and all dependencies I got your workflow working and it's very handy!
EDIT: I think something may be not working ... in the image output of your workflow I am getting "AA AB AC AD AE" in both X and Y axis.
Still, I can't directly use the output of my co-occourence table, since I need to take into account the number of co-occourences and add rows to my table as necessary: i.e. if I am understanding the algorithm correcly, it means converting this:
rowID |
term1 |
term2 |
sentence co-occourence |
row0 |
AA |
BB |
4 |
row1 |
BB |
CC |
2 |
row2 |
AA |
CC |
2 |
into something like this:
rowID |
term1 |
term2 |
row0 |
AA |
BB |
row1 |
AA |
BB |
row2 |
AA |
BB |
row3 |
AA |
BB |
row4 |
BB |
CC |
row5 |
BB |
CC |
row6 |
AA |
CC |
row7 |
AA |
CC |
I guess I will need to do collumn / row filtering, grouping and pivoting right? Thanks for the help!
Hi I am glad that the workflow worked, and yes the data should be displayed like the second table.
Best Regards
Something must be wrong. Only variables from "A" columns (AA, AB, AC, AD, and AE) are displaying in the matrix. Variables from "B" column are not appearing ...