How to create a classifier for data other then Images?


(Rushabh Vasani) #1

As in lesson 8 we can make a ModelData object for Images.Like for Mnist data we can call ImageClassifierData class to wrap the data for training. So if I want to create a Data object for my model like titanic([https://www.kaggle.com/c/titanic/data]) than which Data wrapper should I use??
I have tried several ways but I got error every time.Like I tried to directly add my train and validation data to fit function in the form of a list but I got the error "the function object have no attribute dl_trn".
and so on so what should I do to fit my model.
(sklearn’s LogisticRegression worked for me but it gave score of 0.7877 and i think pytorch will give better score because of multi-layers)
So what should I do to wrap my data for fit function??


(Sam Ariabod) #2

I am not sure I fully understand the question, but I think you are asking about Tabular data functionality (which I think the titanic data is).

https://docs.fast.ai/tutorial.inference.html#Tabular


(Rushabh Vasani) #3

Thanks @sariabod…!
I found it very helpful because I didn’t really knew about this tabular data functionality but I have a question that can we make multi-layers in this tabular_learner like a deep NN?