csv loader and KNN prediction example
notice that the data downloaded and submitted to the kaggle should all be like this:
[ [ 'id', 'age', ... ,'category_n'],
[ 1, 20, ... , 2019.90], .... ,
[100, 30, ... , 2100.90 ] ]
test accuracy around 0.58-0.66 with K in 1,3,5,...,201
