Challenge : How to do Category mapping to new Part Numbers using a training data set

[JohanRanschaert]

Hi all,

I have a list of Part Numbers and the related Category.

Based out of this set of data, which I would call the training data set, I would like to build a model.

This model would then be used to define the Category for new Part Numbers for which I don't know the Category yet.

Where should I start? How is this called? Pattern Matching, Predictive Grouping, Clustering, ...?

Anyone with experience in regards to this topic?

Thx for the input.

Johan

[mceleavey]

HI @JohanRanschaert ,

can you post some data showing what you have and what you're trying to achieve so we can help?

M.

[JohanRanschaert]

Hi,

thx for reaching out.

In attach a sample file.

Regards,

Johan

[JohanRanschaert]

Hi Mister,

thx for reaching out.

In attach a sample file.

Regards,

Johan

[danilang]

Hi @JohanRanschaert 

Since you're attempting to predict a categorical variable with categorical predictors, you should look at the decision tree or forest model tools. 

The difficult part will be normalizing the part numbers.  For instance, in the following two part numbers,  "HA2" is possibly significant, but it's fourth in the one and fifth in the other.   

PAN-PA-5050-GP-HA2-R
PAN-PA-5020-URL2-5YR-HA2

You'll need to develop a scheme to modify part numbers as required to properly align the pieces.

I suppose that contacting the team that mapped your training set and asking them how they did it is out of the question.

Dan

[JohanRanschaert]

Hi Sir,

Thx for the input.

This is the reason why I would like to train a model so we don't need to normalize the sku or cut the sku into pieces.

By the way I tried using Decision Tree and Forest Model.

When using Decision Try, the flow was running for more then 20 hours still at 21%

When using Forest Model, I always get the same errors:

- Error: Forest Model (5): Forest Model: Error in randomForest.default(m, y, ...) :

- Error: Forest Model (5): Forest Model: Execution halted

- Error: Forest Model (5): Forest Model: The R.exe exit code (1) indicated an error.

Regards,

Johan