Subsequent geom_tile call is not zero indexed for x axis.

[joseph-jnl]

Hello, I'm trying to understand why my second call to geom_tile is not zero indexed like the first.

i.e.. The first geom_tile uses zero indexed rows and columns but the second geom_tile call starts at 1 instead for the x axis. The intent is to highlight the tiles at row 0 and column 1 and 3 rather than columns 0 and 2.

Thank you!

[has2k1]

Please post a minimal reproducible example.

[joseph-jnl]

import pandas as pd
from plotnine import ggplot, aes, geom_tile, scale_y_reverse

df1 = pd.DataFrame([[0,1,2], [3,4,5], [6,7,8]])
df1 = df1.reset_index().melt('index')
df1.columns = ['row', 'column', 'value']
df2 = pd.DataFrame({"row": [0, 0], "column": [1, 2], "value": list(range(2))})

(ggplot(df1, aes("column", "row", fill="value")) 
 + geom_tile()
 + geom_tile(data=df2, 
             size=2, fill=None, color="black")
 + scale_y_reverse()
 )

The current output aligns with what I would expect a df2 with column set to [0, 1].

[has2k1]

With the way you create the dataframe, df1["column"] is discrete (dtype=object) while df2["column"] is continuous (dtype=int).

print({k: str(v.dtype) for k, v in df1.items()})
print({k: str(v.dtype) for k, v in df2.items()})

# {'row': 'int64', 'column': 'object', 'value': 'int64'}
# {'row': 'int64', 'column': 'int64', 'value': 'int64'}

So if you do

df1["column"] = df1["column"].astype(int)

It should act as you expect.