Small things that drove me nuts – and how they were resolved… an ever-expanding list, no doubt 🙂
Issue |
Solution |
Source |
| plt.savefig() saves an empty picture or a cut-off picture | plt.savefig() should be placed BEFORE plt.show() to prevent this happening | Stackoverflow solution |
| Pandas prints pie chart labels and a legend which is redundant | To lose the legend: ct.plot.pie(“Submission”, legend = False) | |
| Pandas prints pie chart labels and a legend which is redundant | To lose the labels: ct.plot.pie(“Submission”, labels = None) plt.legend(labels = ct.index) |
Stackoverflow solution |
| df.rename() with inplace = True gives a warning “A value is trying to be set on a copy of a slice from a DataFrame” | Use a workaround: my_data.columns = my_data.columns.str.replace(“OldName”, “NewName”) |
Stackoverflow discussion and solutions |
| plt.imshow() distorts your image according to axis size | There is a parameter to resolve this! plt.imshow(…, aspect = “auto”) | |
| Image retrieved via Google static maps API is poor quality | Use scale = 2 which retrieves e.g. 1280×1280 pixels for a 640×640 image | Google developer documentation |
| get rid of column index levels from pivot | df.columns.droplevel() resolves(…, aspect = “auto”) | Stackoverflow solution |
| save direct to GCS bucket without local storage | Use the StringIO solution – works like a bomb! | Stackoverflow solution |
| pandas list transform to multiple rows | It’s a big snippet – head to stackexchange for the complete solution | Stackoverflow solution |
Sho't left to data science 