SAR item similarity dtype correction

[simonzhaoms]

Description

In sar_singlenode.py, item similarity matrix is converted to the same dtype as col_rating:
https://github.com/microsoft/recommenders/blob/d4181cf1d1df6e71f7e6b202b0875bb3bd54150c/recommenders/models/sar/sar_singlenode.py#L278-L280

When the dtype of col_rating is int, all numbers will be rounded off to the nearest integer which is not expected:

>>> import pandas as pd
>>> import numpy as np
>>> header = {
...     "col_user": "UserId",
...     "col_item": "MovieId",
...     "col_rating": "Rating",
... }
>>> ratings_dict = {
...     header["col_user"]: [1, 1, 2, 3, 3],
...     header["col_item"]: [1, 2, 1, 1, 3],
...     header["col_rating"]: [1, 2, 1, 1, 2],
... }
>>> df = pd.DataFrame(ratings_dict)
>>> df[header["col_rating"]].dtype
dtype('int64')
>>> from recommenders.models.sar import SAR
>>> model = SAR(**header, similarity_type="jaccard")
>>> model.fit(df)
>>> print(model.item_similarity)
[[1 0 0]
 [0 1 0]
 [0 0 1]]

The correct item similarity should be

[[1.         0.33333333 0.33333333]
 [0.33333333 1.         0.        ]
 [0.33333333 0.         1.        ]]

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.

[miguelgfierro]

LGTM