Algos visibility

[miguelgfierro]

Description

I was talking with a colleague that is interested in graph recommenders. I realized that it is not straightforward for people that don't know the repo to know that sometimes there are quickstart notebooks and sometimes deep dive notebooks.

Not sure what is the best way to give visibility to all the content. Maybe we could consider adding the deep dive into the main readme? Other option could make a comment in the deep dive, referencing the quickstart, and viceversa, in the markdown of the notebook.

@anargyri what do you think?

Related Issues

Checklist:

• I have followed the contribution guidelines and code style for this project.
• I have added tests covering my contributions.
• I have updated the documentation accordingly.
• This PR is being made to staging branch and not to main branch.

[anargyri]

How about including both deep dive and quick start for each algorithm in the table?
It seems that increasingly people use this table as a comprehensive list of the notebooks that are available.

[miguelgfierro]

How about including both deep dive and quick start for each algorithm in the table?

Makes sense

[miguelgfierro]

I'm going to put here some options:

Option 1 (Original):

Algorithm	Environment	Type	Description
Alternating Least Squares (ALS)	PySpark	Collaborative Filtering	Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability

Option 2 (add extra column):

Algorithm	Environment	Type	Description	Example
Alternating Least Squares (ALS)	PySpark	Collaborative Filtering	Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability	Quick start / Deep dive

Option 3 (remove environment column and add it in description):

Algorithm	Type	Description	Example
Alternating Least Squares (ALS)	Collaborative Filtering	Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability	Quick start in PySpark / Deep dive in PySpark

Option 3.5 (remove environment column and add it in description):

Algorithm	Type	Description	Example
Alternating Least Squares (ALS)	Collaborative Filtering	Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability	Quick start / Deep dive

Option 4 (option 3 with reordered columns):

Algorithm	Example	Type	Description
Alternating Least Squares (ALS)	Quick start in PySpark / Deep dive in PySpark	Collaborative Filtering	Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability

Option 5 (variation of 4 removing the word "Python" to save space):

Algorithm	Python Notebook	Type	Description
Alternating Least Squares (ALS)	Quick start in Spark / Deep dive in Spark	Collaborative Filtering	Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability

[laserprec]

I'm going to put here some options:

I would vote for option 3, as our eyes are naturally drawn to either ends of the page. I am not sure how important it is to provide the environment information to users. As a first time user, I didn't find the environment info that much helpful because I would try out the algorithm regardless of the environment anyways if it is interesting 😄.
I would prefer

without the clutter.

[laserprec]

My preference (modified option 3) 😄

Option 3.5 (remove environment column and add it in description):

Algorithm	Type	Description	Example
Alternating Least Squares (ALS)	Collaborative Filtering	Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability	Quick start / Deep dive

[anargyri]

I would vote for option 2 or option 3.5. I think each algorithm we have maps 1-1 to an environment, so both notebooks use the same environment. You can specify it in a separate column or omit it. I tend towards including the env info, because e.g. some people want to try the repo on a machine without GPU and it is convenient to see which methods may be slower because of this.

[miguelgfierro]

@laserprec I added column 3.5 in the list.

@yueguoguo @loomlike @gramhagen @wutaomsft any preference?

[miguelgfierro]

ok, I'll implement 3.5 as suggested by @anargyri and @laserprec

[miguelgfierro]

@anargyri @laserprec, please feel free to review again. If you find a typo or small thing that need to be change, as in other times, feel free to change it directly

[codecov-commenter]

Codecov Report

Merging #1542 (69cbc64) into staging (2306b2b) will decrease coverage by 0.23%.
The diff coverage is n/a.

❗ Current head 69cbc64 differs from pull request most recent head 7ee14fb. Consider uploading reports for the commit 7ee14fb to get more accurate results

@@             Coverage Diff             @@
##           staging    #1542      +/-   ##
===========================================
- Coverage    62.36%   62.13%   -0.24%     
===========================================
  Files           85       84       -1     
  Lines         8503     8437      -66     
===========================================
- Hits          5303     5242      -61     
+ Misses        3200     3195       -5     

Flag	Coverage Δ
pr-gate	62.13% <ø> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
recommenders/datasets/movielens.py	68.44% <0.00%> (-5.47%)	⬇️
recommenders/models/newsrec/newsrec_utils.py	66.66% <0.00%> (-3.34%)	⬇️
recommenders/models/geoimc/geoimc_data.py	39.13% <0.00%> (-2.54%)	⬇️
...ommenders/models/deeprec/io/sequential_iterator.py	14.34% <0.00%> (-1.51%)	⬇️
recommenders/models/lightfm/lightfm_utils.py	62.50% <0.00%> (-1.02%)	⬇️
recommenders/models/geoimc/geoimc_algorithm.py	85.50% <0.00%> (-0.80%)	⬇️
...ders/models/deeprec/models/sequential/sum_cells.py	44.75% <0.00%> (-0.77%)	⬇️
recommenders/models/surprise/surprise_utils.py	79.31% <0.00%> (-0.69%)	⬇️
recommenders/models/newsrec/models/base_model.py	30.48% <0.00%> (-0.43%)	⬇️
recommenders/models/newsrec/models/layers.py	52.99% <0.00%> (-0.40%)	⬇️
... and 29 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 448982d...7ee14fb. Read the comment docs.

[miguelgfierro]

@laserprec @anargyri when you are available, can you please review again?

[laserprec]

@laserprec @anargyri when you are available, can you please review again?

I double checked all the links again. Looks good!