I want to predict some typo shortcuts.
For example:
8 in. micrometer has to be predicted as 8 inch micrometer 9 lbs Bag - 9 pounds bag 10" scale - 10 inch scale 10 no. - 10 numbers 77 mm length - 77 millimeter length and so on. I already created a small dataset of 80 lines. But, I need a large training dataset of english words and their shortcuts also, i am using RandomForest algorithm for predicting. I wanted to know which algorithm is better for text normalization and I wanted to know how much test size we can have because i faced issues of accuracy being low and high when i am changing the test size.
