Options
Is RapidMiner able to solve my problem?
eifel_wolf
Member Posts: 7 
Contributor II
Hello,
even I'm interessted in Data Mining and I also know about RapidMiner since a long time I didn't manage to get really started ... So I'm an absolutely Newbie to this.
Actually there is situation in my work which I hope RapidMiner will be able to help us, but I'm not sure and I was not able to find an example for a similar situation. So I will try to ask the community.
We are going to monitor customers and their sales in a long term. Unfortunately we will now get a change of key id's, this means all of our customers will get new id's within the next months. We are not able to get a bridging file to match the old id of a customer with the new one, circumstances will not allow us to get or create such a thing out of the sources. The only thing we have available are items like age, gender and member of some groups - and of course the sales we are monitoring.
My idea is to use RapidMiner to create such a bridging list on our own ... it means getting the "fingerprint" of a customer out of the data we have available and find the matching pairs between old and new id.
What do you think, are there ways to do this with RapidMiner? And if there are ways - can you give me some links to examples or good search keywords to find similar things?
Regards
Wolf
even I'm interessted in Data Mining and I also know about RapidMiner since a long time I didn't manage to get really started ... So I'm an absolutely Newbie to this.
Actually there is situation in my work which I hope RapidMiner will be able to help us, but I'm not sure and I was not able to find an example for a similar situation. So I will try to ask the community.
We are going to monitor customers and their sales in a long term. Unfortunately we will now get a change of key id's, this means all of our customers will get new id's within the next months. We are not able to get a bridging file to match the old id of a customer with the new one, circumstances will not allow us to get or create such a thing out of the sources. The only thing we have available are items like age, gender and member of some groups - and of course the sales we are monitoring.
My idea is to use RapidMiner to create such a bridging list on our own ... it means getting the "fingerprint" of a customer out of the data we have available and find the matching pairs between old and new id.
What do you think, are there ways to do this with RapidMiner? And if there are ways - can you give me some links to examples or good search keywords to find similar things?
Regards
Wolf
0
Answers
with Data Mining it is possible to train models that are able to find groups with similar customers. But unfortunately I think it is not possible to
get an accurate enough model to predict the correct ID mapping you are looking for.
Best,
Nils
thank you for the quick feedback!
About your concerns: We know that we wont be able to get a 100 % perfect matching table by this, but we hope that we will be able to get a good match about some of the id's.
We are in a special environment and one important point in there is that we are looking for id's having the same kind of sales (same products or products of the same categorie, combination of produkts/categories) in a long term, sales are generated in an individual interval. Additionally we will be able to define some "must match" keys to make the assignment more easy by making the amount of potential matches smaller.
And, at all, in the end we can check if the sales go seamless from old to new id, which means no double sales and no gap of sales during change process.
From our knowledge about the data it must be possible to find good matches for a big part of the id's, but we are not sure how to do this. It's the huge amount of data at all and even the huge amount of possible relevant criterias which makes it difficult to create a match algorithm. That's why we think data mining tools may help us ...
Regards
Wolf
with this kind of background knowledge it might be possible to solve your problem.
But I can't guarantee it without haven a look on the data myself.
At least you can try by importing the data into RapidMiner and try to build clusters.
Best,
Nils
thank you for your feedback! I will start with the cluster-thing you've mentioned ... At all I first have to get a little bit more famillar with data mining in generall and with RapidMiner it self too ;-)
Regards
Wolf