One of the challenges when working with categorical data in pandas and Holoviews can be ordering appearance in a plot. To get around this, if you sort the pandas dataframe prior to sending the data to your Holoviews boxplot you can avoid the issue.

Source for example: https://stackoverflow.com/a/63508053

# Sort the df by Treatment for later ordering.
def trt_sorter(column):
     """Sort function"""
     trt_order = ['1_naive','2_naive_wc','2_even-shuf_wc','2_flip1_wc','2_flip50_wc']
     correspondence = {trt: order for order, trt in enumerate(trt_order)}
     return column.map(correspondence)
 Remove the "10000" from treatments.
 df['Treatment'] = df['Treatment'].map(lambda x: x.rstrip('_10000'))
 df = df.sort_values(by="Treatment", key=trt_sorter)