mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Pat Ferrel <...@occamsmachete.com>
Subject Re: FP-Growth deprecated
Date Tue, 30 Jun 2015 17:30:35 GMT
Yes, that is one of the most common uses of fp-growth.

Instead a better way to look at this might be using cooccurrence. If you collect the items
bought in every shopping cart, one row per cart and boolean value for every item bought, then
perform mahout’s spark-itemsimilarity this will produce a matrix row keyed by items and
rows consisting of items most often bought with the key item. The Driver output is sorted
so the most common items are the first in the list.

Then for a given shopping cart with items in it, you have a knn problem where you want to
find items that were bought with the same list in the current shopping cart. This is done
by indexing the output of spark-itemsimilarity with a search engine and querying with the
current contents of the cart.
  
This is exactly how we do recommendations for individuals but instead of training with user-item
interactions and querying with user history we are training with shopping carts and querying
with the current contents.

See this reference and replace every reference to a user with some shopping cart id: http://mahout.apache.org/users/algorithms/recommender-overview.html
http://mahout.apache.org/users/algorithms/intro-cooccurrence-spark.html

On Jun 30, 2015, at 12:36 AM, guo weizhan <guo.weizhan@gmail.com> wrote:

We want FP-Growth to do the Market Basket Analysis, is there any other
algorithm we can use?

Thanks,
Guo

2015-06-22 8:37 GMT+08:00 Pat Ferrel <pat@occamsmachete.com>:

> What is your application?
> 
> On Jun 17, 2015, at 7:06 AM, guo weizhan <guo.weizhan@gmail.com> wrote:
> 
> Hi All,
> 
> I found the FP-Growth was deprecated since 0.8, but we want this algorithm
> to do the association analysis. Do I have to use the old version or  Is
> there any other association analysis I can use in the lastest version?
> 
> 
> Thanks,
> Guo
> 
> 


Mime
View raw message