python - Pandas Join on String Datatype

Question

I am trying to join two pandas dataframes on an id field which is a string uuid. I get a Value error:

ValueError: You are trying to merge on object and int64 columns. If you wish to proceed you should use pd.concat

The code is below. I am trying to convert the fields to string as per Trying to merge 2 dataframes but get ValueError but the error remains. Note that pdf is coming from a spark dataframe.toPandas() while outputsPdf is created from a dictionary.

pdf.id = pdf.id.apply(str)
outputsPdf.id = outputsPdf.id.apply(str)
inOutPdf = pdf.join(outputsPdf, on='id', how='left', rsuffix='fs')

pdf.dtypes
id         object
time      float64
height    float32
dtype: object

outputsPdf.dtypes
id         object
labels    float64
dtype: object

How can I debug this? Full Traceback:

ValueError                                Traceback (most recent call last)
<ipython-input-13-deb429dde9ad> in <module>()
     61 pdf['id'] = pdf['id'].astype(str)
     62 outputsPdf['id'] = outputsPdf['id'].astype(str)
---> 63 inOutPdf = pdf.join(outputsPdf, on=['id'], how='left', rsuffix='fs')
     64 
     65 # idSparkDf = spark.createDataFrame(idPandasDf, schema=StructType([StructField('id', StringType(), True),

~/miniconda3/lib/python3.6/site-packages/pandas/core/frame.py in join(self, other, on, how, lsuffix, rsuffix, sort)
   6334         # For SparseDataFrame's benefit
   6335         return self._join_compat(other, on=on, how=how, lsuffix=lsuffix,
-> 6336                                  rsuffix=rsuffix, sort=sort)
   6337 
   6338     def _join_compat(self, other, on=None, how='left', lsuffix='', rsuffix='',

~/miniconda3/lib/python3.6/site-packages/pandas/core/frame.py in _join_compat(self, other, on, how, lsuffix, rsuffix, sort)
   6349             return merge(self, other, left_on=on, how=how,
   6350                          left_index=on is None, right_index=True,
-> 6351                          suffixes=(lsuffix, rsuffix), sort=sort)
   6352         else:
   6353             if on is not None:

~/miniconda3/lib/python3.6/site-packages/pandas/core/reshape/merge.py in merge(left, right, how, on, left_on, right_on, left_index, right_index, sort, suffixes, copy, indicator, validate)
     59                          right_index=right_index, sort=sort, suffixes=suffixes,
     60                          copy=copy, indicator=indicator,
---> 61                          validate=validate)
     62     return op.get_result()
     63 

~/miniconda3/lib/python3.6/site-packages/pandas/core/reshape/merge.py in __init__(self, left, right, how, on, left_on, right_on, axis, left_index, right_index, sort, suffixes, copy, indicator, validate)
    553         # validate the merge keys dtypes. We may need to coerce
    554         # to avoid incompat dtypes
--> 555         self._maybe_coerce_merge_keys()
    556 
    557         # If argument passed to validate,

~/miniconda3/lib/python3.6/site-packages/pandas/core/reshape/merge.py in _maybe_coerce_merge_keys(self)
    984             elif (not is_numeric_dtype(lk)
    985                     and (is_numeric_dtype(rk) and not is_bool_dtype(rk))):
--> 986                 raise ValueError(msg)
    987             elif is_datetimelike(lk) and not is_datetimelike(rk):
    988                 raise ValueError(msg)

Answer 1

The on parameter only applies to the calling DataFrame!

on: Column or index level name(s) in the caller to join on the index in other, otherwise joins index-on-index.

Though you specify on='id' it will use the 'id' in pdf, which is an object and attempt to join that with the index of outputsPdf, which takes integer values.

If you need to join on non-index columns across two DataFrames you can either set them to the index, or you must use merge as the on paremeter in pd.merge applies to both DataFrames.

---

Example

import pandas as pd

df1 = pd.DataFrame({'id': ['1', 'True', '4'], 'vals': [10, 11, 12]})
df2 = df1.copy()

df1.join(df2, on='id', how='left', rsuffix='_fs')

ValueError: You are trying to merge on object and int64 columns. If you wish to proceed you should use pd.concat

On the other hand, these work:

df1.set_index('id').join(df2.set_index('id'), how='left', rsuffix='_fs').reset_index()
#     id  vals  vals_fs
#0     1    10       10
#1  True    11       11
#2     4    12       12

df1.merge(df2, on='id', how='left', suffixes=['', '_fs'])
#     id  vals  vals_fs
#0     1    10       10
#1  True    11       11
#2     4    12       12