How to create blank tabular model?

Part 1 (2019)
1

Hello,

I am trying to create a tabular learner model with no data then I will assign the data later
Is there any way to do this?
Something like this

learn = tabular_learner( NONE , layers=[100,200], metrics=accuracy)

Thankyou

2

Taking into account that a databunch instance is compulsory in order to create a Learner I don`t think it is posible…

With what purpose do you want to do that in that specific way?

3

Thanks for the reply

I am trying to create a class on top of FastAI library to use the simplify function
like sklearn or keras where you can fit for example

model = tabular_learner( NONE , layers=[100,200], metrics=accuracy)
model.fit(X_train,Y_train)

or some quick way to predict on new data using just 1 line where most of the
databunch will be handled inside the class like

Y_prd = model.predict(X_train)

or simple calculate accuracy from defined dataset

model.score(X_train,Y_train)
model.score(X_test,Y_test)

This is my personal project just to simplify my workflow,
If you heard of any library on top of FastAI that somebody has worked on, Please let me know :slight_smile: