1
$\begingroup$

I have a problem to name data processing step.

  1. I have an attribute that contain string or null. I want to change the record of an attribute to 0 if null and 1 if not null. What preprocessing step name?
  2. I have an attribute that contain 200 tweets of a twitter user. I make a new attribute that describe URL ratio in 200 tweets (number of URL in 200 tweet / 200). is it aggregation or feature creation or else?
$\endgroup$
1
$\begingroup$

You are looking for names to attribute to the two items listed? For (1) I would just call it a transformation as it is a straight mapping with no change in distribution of the feature. When working with data, make sure you make copies of your data transformation and do not alter the original data set.

For (2), since it is a single number per group, where group here is the full data set I would call it an aggregation. Likewise if you did a similar calculation per user. If however, you computed a new value from existing features for each record, this would be feature generation or creation.

$\endgroup$
  • $\begingroup$ additional question. i'm crawling data from twitter. as we know, twitter will return data with json format. i do some process to reduce data, like i do when i get status from user that return group of tweet (200 tweet). i just calculate URL number after get data from one user, and make URL number feature. can i still named it as crawling when it return not original data from twitter? $\endgroup$ – dionajie Dec 10 '15 at 1:56
  • $\begingroup$ I would call the first step as dummy encoding though. (Even, transformation is also close :) ) $\endgroup$ – Dawny33 Dec 10 '15 at 3:31
1
$\begingroup$

I have attribute that contain string or null. i want to change the record of attribute to 0 if null and 1 if not null. What preprocessing step name?

This step is dummy coding.

i have attribute that contain 200 tweet of twitter user. i make new attribute that describe URL ratio in 200 tweet (number of URL in 200 tweet / 200). is it aggregation or feature creation or else?

Feature aggregation

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.