Not define a reducer and cleaner method

Matminer
1

Hi,

I would like to create a MatPipe without the necessity to define a reducer and cleaner method. Is there a way to do so?

Thanks!

2

Hey Osmar,

Currently, no. This is a good issue I had not considered: The matpipe is just a way to link together multiple modular data frame transformers. I’ll add it as a issue todo to take of this.

However (in the meantime), each part of the automatminer pipeline can be used with essentially the same syntax as MatPipe (fit/transform/predict operations). You can just use the underlying parts by themselves in whatever order you want (not recommended but if you know what you’re doing go for it). They accept a df as input and return a df as output, just like MatPipe. For example, initialize an AutoFeaturizer and TPOTAdaptor and run the fit/transform actions manually. Just make sure you are using fit/transform/predict correctly and things should be fine.

For example

af = AutoFeaturizer()

tpot = TPOTAdaptor()

training_df = af.fit_transform(training_df, “my_target_col”)

tpot.fit(training_df, “my_target_col”)

prediction_df = af.transform(prediction_df, “my_target_col”)

tpot.predict(prediction_df, “my_target_col”)

Check out the fit/transform methods of MatPipe for more tips on how to use the underlying AutoFeaturizer/MLAdaptor pieces.

Thanks,

Alex

···

On Thu, Jul 18, 2019 at 2:23 PM Osmar [email protected] wrote:

Hi,

I would like to create a MatPipe without the necessity to define a reducer and cleaner method. Is there a way to do so?

Thanks!

–

You received this message because you are subscribed to the Google Groups “matminer” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To view this discussion on the web visit https://groups.google.com/d/msgid/matminer/e0baeb9c-31f6-4321-9461-bacf4adf90e1%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.