Hi Jorge,
How did you install pyarrow 4.0.0? The error you show typically points to
an installation issue (eg built with a wrong numpy)
Best,
Joris
On Tue, 27 Apr 2021 at 16:47, Jorge Alarcon <jorge.alar...@maccresi.com>
wrote:
> Hi everybody,
>
>
>
> Please, there is an issue with pyarrow (version 4.0.0) when you try to
> write a parquet with your engine. It is not possible to write a parquet
> from a pandas df when it includes non str columns (datetime64, float64,
> int64…)
>
>
>
> Example:
>
>
>
> df = pd.DataFrame({'A':[1, 2, 3], 'B':['a', 'b', 'c']})
>
> df.to_parquet('example.parquet', engine='pyarrow') #Not working
>
> *ArrowTypeError*: ('Did not pass numpy.dtype object', 'Conversion failed
> for column InternalId with type float64')
>
>
>
> df['A'] = df['A'].astype(str)
>
> df.to_parquet('example.parquet', engine='pyarrow') #Working
>
>
>
> Best!
>
>
>
> *Jorge Alarcon*
>
> *Senior Data Analytics Specialist*
>
>
>
> Mail: jorge.alar...@maccresi.com
>
> Telf: +34 683541389
>
> 28020 Madrid
>
>
>
>
>
