1
$\begingroup$

Ok, so I am trying to classify a rather large data set where the training set has some peculiar issues...

  • There are a different number of features available for each row. For example, I might have 10 data points to work with for one prediction and only 3 for another. I would want to somehow null the empty data points because they aren't necessarily 0. We just don't have the data.

  • Each feature is a triplet of numbers (A,B,C), (A1,B1,C1), (A2,B2,C2). Should I separate these into individual features or is there a way to keep them grouped?

  • There are a few features/variables which all rows will have.

Is there preferred methodology?

| Y | v0        | v1        | v2      | v3     | v4 | v5 | v6 | v7 | v8 | v9 | A | B | C  |
|---|-----------|-----------|---------|--------|----|----|----|----|----|----|---|---|----|
| A | 15,0,7    | 18,0,2    | 23,0,0  | 21,0,3 |    |    |    |    |    |    | 1 | 0 | 5  |
| B | 15,.05,12 | 14,.05,12 | 9,.3,45 |        |    |    |    |    |    |    | 2 | 2 | 12 |
| A | 18,0,2    | 23,.03,0  |         |        |    |    |    |    |    |    | 0 | 1 | 3  |
| A | 23,0,0    | 18,0.07,2 | 15,0,7  |        |    |    |    |    |    |    | 3 | 1 | 1  |
| C | 4,.08,212 |           |         |        |    |    |    |    |    |    | 6 | 4 | 4  |
| B | 14,.05,12 | 9,.3,45   |         |        |    |    |    |    |    |    | 1 | 3 | 9  |
$\endgroup$
0
$\begingroup$

The simple approach is:

  • To use a learning algorithm which can handle missing values. You should be careful about whether these missing values occur randomly or not though, as they could cause the model to be biased.
  • Triplets can be separated into individual features, there's no simple way to process them differently anyway.
$\endgroup$

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.