I am creating a neural network to denoise music.
The input to the model is an array that is scaled from 0 to 1. This is achieved using sklearn MinMaxScaler. The original range of the data is from -1 to 1. The output of the model is also an array scaled from 0 to 1.
I cannot however scale the data back to -1 to 1 when predicting information.
My code is similar to:
data = load(data_path)
scaler = MinMaxScaler(feature_range = (0,1))
data = data.reshape(-1,1)
data = scaler.fit_transform(data)
model = load_model(model_path)
predicted_data = model.predict(data)
predicted_data = scaler.inverse_transform(predicted_data)
I however receive the error:
This MinMaxScaler instance is not fitted yet. Call 'fit' with appropriate arguments before using this method.
The data however is already fitted and I do not want to fit it again.
Why exactly am I getting this error? Shouldn't MinMaxScaler still be able to do inverse_transform on unfitted data?
Are there any suggestions around this error?
The error says it all, you need to call the fit and transform method separately, and not just fit_transform.
data = load(data_path)
data = data.reshape(-1, 1)
scaler = MinMaxScaler(feature_range = (0,1)).fit(data)
data = scaler.transform(data)
model = load_model(model_path)
predicted_data = model.predict(data)
predicted_data = scaler.inverse_transform(predicted_data)