pd.merge 生成新的列名

Question

Merging 2 dataframes which have some common and some different column names. Results in new column names which are in neither but combine the string of names in each.

2 dataframes:

df.columns has among others 'particle', 'frame', 'x old', 'y old' corrected_traj.columns has 'particle', 'frame', 'x', 'y'

neither dataframe has 'frame_x' or 'frame_y'.

yet when I try to merge I end up with no column named 'frame' but 2 new columns 'frame_x' and 'frame_y'

Neither dataframes index is currently named although they are linked to the frame number. I have been trying to avoid an error associated where the index and a column number have the same name. Hence some of the code trying to remove index names etc. Not sure whether this is relevant so have included.

The dataframes are being produced by functions from trackpy but I think the issue is related to the pd.merge.

The overall aim is to subtract the mean drift of some particles from the motion of the particles. I want to move the old x and y to 'x old' and 'y old' and put the corrected values in 'x' and 'y'

---

drift = tp.motion.compute_drift(df)
corrected_traj = tp.motion.subtract_drift(df[['frame','x','y','particle']].copy(), drift)

df['x old'] = df['x'].copy()
df['y old'] = df['y'].copy()


df = df.drop(columns=['x','y'])
corrected_traj.index.name=None

df = pd.merge(df, corrected_traj,
                  on='particle')

---

Apologies I've tried indenting and hitting the code thing but can't seem to get it to markup correctly

I was expecting a dataframe df with 'x','y','frame','particle','x old', 'y old'.

Instead I'm getting 'x','y', 'frame_x', 'frame_y', 'x old','y old', 'particle'

The contents of 'frame x and 'frame y' do seem to be the frame number values.

Answer 1

If you want to have a dataframe df with 'x','y','frame','particle','x old', 'y old', then you should merge as below to incorporate both particle and frame columns in joining scope. Otherwise they will be treated as columns to be joined based on parameters passed to "on" and thus be treated as frame_x, frame_y to identify them separately.

df = pd.merge(df, corrected_traj, on=['particle', 'frame'])