TabularDataLoaders.from_df why would we need path?

fastai
1

if we loading from a df why would we need path in fastai - Tabular data , we have the df, it has all the data.

I’ve tried with and without the path both return the same error. I’m working on Windows
path = Path.cwd()
dls = TabularDataLoaders.from_df(df2, procs=procs, cat_names=cat_names, cont_names=cont_names, y_names=“BetResult”, bs=64)

returns
error :can’t convert np.ndarray of type numpy.object_. The only supported types are: float64, which traces back to return to.dataloaders(path=path, **kwargs)

I’m going to try another way then switch to colab or something.

2

try and install WSL (windows subsystem for linux) and setup for working directory there

windows and linux handle path differently which is incompatible so please switch to WSL
u can find the direction to do the same in livecoding 01

1 Like