Attempting Getting Started Python script in Release 0.6.0 Lenskit Documentation

[Bill Noffsinger]

I assume the getting started document covers a Python script in Jupyter Notebook.  I have Anaconda Navigator 1.9.6 and have the latest Python installed, When entering the script I receive an error on the %matplotlib inline 

statement. I can comment that out and continue since it is not used until later in the script. 

I receive this result early in the script. The data values look correct but the NaN values suggest an error to me.

user	item	rating	timestamp
0	196\t242\t3\t881250949	NaN	NaN	NaN
1	186\t302\t3\t891717742	NaN	NaN	NaN
2	22\t377\t1\t878887116	NaN	NaN	NaN
3	244\t51\t2\t880606923	NaN	NaN	NaN
4	166\t346\t1\t886397596	NaN	NaN	NaN

After running the script to this point:

# -*- coding: utf-8 -*-

"""

Created on Thu Feb 21 14:09:48 2019

@author: william.noffsinger

"""

from lenskit import batch, topn, util

from lenskit import crossfold as xf

from lenskit.algorithms import Recommender

from lenskit.algorithms import als

from lenskit.algorithms import item_knn as knn

from lenskit import topn

import numpy

import pandas as pd

#%matplotlib inline

ratings = pd.read_csv('C:/Users/william.noffsinger/Lenskit/ml-100k/ml-100k/u.data', sep='t', names=['user', 'item', 'rating', 'timestamp']) 

ratings.head()

algo_ii = knn.ItemItem(20)

algo_als = als.BiasedMF(50)

def eval(aname, algo, train, test):

    fittable = util.clone(algo)

    fittable = Recommender.adapt(fittable)

    fittable.fit(train)

    users = test.user.unique()

    recs = batch.recommend(fittable, users, 100)

    recs['Algorithm'] = aname

    return recs

all_recs = []

test_data = []

for train, test in xf.partition_users(ratings[['user', 'item', 'rating']], 5, xf.SampleFrac(0.2)):

    test_data.append(test)

    all_recs.append(eval('ItemItem', algo_ii, train, test))

    all_recs.append(eval('ALS', algo_als, train, test))

    #####

    all_recs = pd.concat(all_recs, ignore_index=True)

    all_recs.head()    

    

I receive an error: ValueError: zero-size array to reduction operation maximum which has no identity

and looking at the Variable Explorer in Spyder  I see the all_recs list size is zero and the test_data list size is 1. 

I assume I have the correct environment for Python, Anaconda, etc. 

Any tips would be appreciated.

[Michael Ekstrand]

Bill,

It looks like your 'read_csv' call doesn't have the right 'sep': it should be "sep='\t'", not "sep='t'". This can be seen in the dump of the data table: the 'item' column has tab-separated numbers in it.

Hope that helps and clears it up! Yes, this is a Jupyter notebook.

- Michael

--
You received this message because you are subscribed to the Google Groups "LensKit Recommender Toolkit Development and Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lenskit-recsy...@googlegroups.com.
To post to this group, send email to lenskit...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lenskit-recsys/9df4546a-e1a4-481b-8907-82624340287a%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Michael D. Ekstrand — michael...@boisestate.edu — https://md.ekstrandom.net
Assistant Professor, Dept. of Computer Science, Boise State University

People and Information Research Team (PIReT) — http://coen.boisestate.edu/piret/

[Bill Noffsinger]

Yes, I fixed my typos and the script ran fine - thanks. Now I will study the docs and code and start making some mods as I learn more of the details. Python (Spyder) still gives me a syntax error on the matplotlib but I will move on.

matplotlib.axes._subplots.AxesSubplot at 0x7f03842f8860>     gets a syntax error in Spyder

Many thanks!

---

From: lenskit...@googlegroups.com <lenskit...@googlegroups.com> on behalf of Michael Ekstrand <michael...@boisestate.edu>
Sent: Wednesday, March 13, 2019 12:06:22 PM
To: Bill Noffsinger
Cc: LensKit Recommender Toolkit Development and Support
Subject: Re: Attempting Getting Started Python script in Release 0.6.0 Lenskit Documentation

 

NSU Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know that the content is safe.

To view this discussion on the web visit https://groups.google.com/d/msgid/lenskit-recsys/CABSXaUtDvMFjPOTZdXvg03NcYZW4biO1LmSUwR9VeQMGTePvhg%40mail.gmail.com.

[Bill Noffsinger]

Doe the current LKPY Lenskit have an algorithm implementation for a Restricted Boltzmann Machine, Recurrent Neural Network or other deep learning implementation? 

[Michael Ekstrand]

Not currently.

We would welcome either an implementation or a bridge to an implementation in another library. Bridges are often preferable, to decrease LensKit code maintenance.

- Michael

--
You received this message because you are subscribed to the Google Groups "LensKit Recommender Toolkit Development and Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lenskit-recsy...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/lenskit-recsys/0ea06c94-b8c4-4251-834c-b6a45397229e%40googlegroups.com.

--

Michael D. Ekstrand — michael...@boisestate.edu — https://md.ekstrandom.net
Assistant Professor, Dept. of Computer Science, Boise State University

People and Information Research Team (PIReT) — http://coen.boisestate.edu/piret/

I may send mail outside of working hours; I do not expect you to. He/him.

[Bill Noffsinger]

I'm experimenting with the Getting Started script in LKPY: 

I wish to display RMSE and MAE metrics. 

I have imported  the predict module as in

     from lenskit.metrics import predict                         no errors or warnings are produced

Next I coded these two lines per page 16 in the LKPY doc:

      preds = predict(algo, pairs)

      user_rmse = preds.groupby('user').apply(lambda df: rmse(df.prediction, df.rating))

Python then says that algo, pairs, rmse are undefined  ??

On Wednesday, March 13, 2019 at 11:34:49 AM UTC-4, Bill Noffsinger wrote:

[Michael Ekstrand]

Bill,

That example is not complete - it assumes you have variables 'algo' and 'pairs' that contain your algorithm and your test data.

I will update the example to be runnable.

- Michael

--
You received this message because you are subscribed to the Google Groups "LensKit Recommender Toolkit Development and Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lenskit-recsy...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/lenskit-recsys/09cf4caa-5036-428a-ac51-df473965fc2f%40googlegroups.com.