Problems with the RAM

SimonK

Hi,

I work with 23 attributes with about 6.5 million values each. When using different operators, the memory of my laptop is often not enough.

Is there an operator to remove rows in which the values hardly change or to reduce the size of the data set?

BalazsBaranyRM

Hi SimonK,

for attributes that don't change a lot you would use "Remove useless attributes". This drops attributes that have little variation in them.

For rows you need a way to define what you mean by "hardly change" if you want to systematically remove them. But then you would probably change the data distribution.

If you have so much data, sampling is a good way to reduce the data size. You could try different sampling sizes. At some point, models don't change anymore even when throwing more data at them.

Regards,
Balázs