Date format changed using seaborn pointplot

Question

When I use seaborn to draw a pointplot, the date on the x-axis changes to this strange format 2020-01-06T00:00:00.000000000. The dates from the 'Current Year Week Ending' column are converted to DateTime object before drawing this figure. My other graph (lineplot) uses similar inputs and format, but it doesn't have this issue, the dates are like 2020-01-06.

Does anyone know how to fix this problem?

*The total_us dataframe and sub dataframe I use in my plots are subsets of the same dataset. And I converted the date on that big dataset before creating those two subsets. So the value in 'Current Year Week Ending' column of total_us and sub should have the same DateTime format.

#create a pointplot to capture the variability                             
plt.figure(figsize = (8, 6))                                                  
sns.pointplot(x = 'Current Year Week Ending',                                               
              y = 'ASP Current Year', 
              hue ='Type', 
              data = sub, 
              markers=["o", "x"],
              linestyles=["-", "--"])                                
plt.xticks(rotation=45, horizontalalignment='right', fontweight='light', fontsize='medium')
plt.ticklabel_format(style='plain', axis='y')

#create a lineplot 
plt.figure(figsize=(10,7))
sns.lineplot(x='Current Year Week Ending', 
             y='Total Bulk and Bags Units', 
             hue='Type', 
             data = total_us);
plt.xticks(rotation=45, horizontalalignment='right', fontweight='light', fontsize='medium')
plt.ticklabel_format(style='plain', axis='y')

this is what sub looks like, so the values in 'Current Year Week Ending' column are like "2020-01-06". I don't know why it changes when I draw a pointplot.

user_5327068 · Answer

I think your date column may just include these trailing zeros, and seaborn deals with it differently depending on how much room there is on the plot. So try the following for your date column:

sub.iloc[:,1] = sub.iloc[:,1].dt.strftime('Y/%m/%d')

If for some reason that doesn't fix the problem (perhaps seaborn point plots and line plots do something different under the hood), then a possible workaround would be to split the text 2020-01-06T00:00:00.000000000 at the letter 'T' and throw out the trailing zeros after. We can use the fact that sns.pointplot returns a matplotlib.Axes object (from the documentation), setting the variable ax equal to your point plot:

ax = sns.pointplot(x = 'Current Year Week Ending',                                               
          y = 'ASP Current Year', 
          hue ='Type', 
          data = sub, 
          markers=["o", "x"],
          linestyles=["-", "--"]) 
ax.set_xticklabels([date_text.get_text().split("T")[0] for date_text in ax.get_xticklabels()])

Date format changed using seaborn pointplot

1 Answers