Scikit-learn: getting same result on all rows when reusing the model
Question:
So I want to figure out some GDP numbers from a country’s GDP primary industry. The earliest data does not have any GDP values so I have trained a model with newer data. My plan is to use that trained model to guess older data.
I then fed new data to the model (the older data) but the model predicts the same number value for all the years!
What am I doing wrong?
PS. I only started with ML so apologies for messy code/ml technique 🙁
EDIT: FIXED. The new data needed to be scaled too 🙂
Answers:
I believe you need to call sc.transform on X1 as well. Otherwise, the scale of the features would be off, and the predictions become erroneous too.
So I want to figure out some GDP numbers from a country’s GDP primary industry. The earliest data does not have any GDP values so I have trained a model with newer data. My plan is to use that trained model to guess older data.
I then fed new data to the model (the older data) but the model predicts the same number value for all the years!
What am I doing wrong?
PS. I only started with ML so apologies for messy code/ml technique 🙁
EDIT: FIXED. The new data needed to be scaled too 🙂
I believe you need to call sc.transform on X1 as well. Otherwise, the scale of the features would be off, and the predictions become erroneous too.