I'm trying to do the very simple/standard task of presenting points (and then polygons created from them) as a graduated colour style, however many points are missing, evinced from the lowest number being higher than numbers I know to exist in the source data csv. I've checked and the csv field is read in as double so that's not the issue. In the original there are 857 rows; when graduated there are about 24. Same in V3.0.2 & 2.18.20. Same (with different lower bounds) for 2 csv files. Running colour conditional formatting on that column/field in excel works fine, data are distributed around well. You can see that compared with single symbol (red, all points), graduated has hardly any.

graduated vs single symbol

When reopening the project in v2.19.20 the range is initially messed up, but can be re-run to make it work, but the contracted range problem remains.

2.18.20 strangeness

What is causing this?

I don't think I'm doing anything differently to usual - this feels like a weird bug.

1 Answer 1


Ok, turns out I imported the csvs with "discard empty FIELDS" left on as is default. Because the column of data I needed was the last one, and the intermediate columns included blank entries, inexplicably this means that QGis slides all data left, such that the only remaining data in the rightmost column will be for those rows which have no blank values. In my opinion this is clearly not what one would expect given that the text refers to empty FIELDS not cells, and all other references to fields in QGIs refer to the columns of data. Further, I can envision why one might wish to exclude ROWS which contain any empty values, but what possible benefit is there for behaviour which slides data from column (e.g.) "weight" into "year" or similar?

  • 1
    That sounds like a serious bug. Please make a bug report to save other QGIS users from having their data corrupted in this way. www.issues.qgis.org
    – csk
    May 31, 2018 at 20:01
  • Already done amigo: issues.qgis.org/issues/19071
    – dez93_2000
    May 31, 2018 at 22:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.