supermemo2 3.0.0

Implemented the SM-2 algorithm for spaced repetition learning.

SuperMemo2

A package that implemented the spaced repetition algorithm SM-2 for you to quickly calculate your next review date for whatever you are learning.

📌  Note: The algorithm SM-2 doesn't equal to the computer implementation SuperMemo2. In fact, the 3 earliest implementations (SuperMemo1, SuperMemo2 and SuperMemo3) all used algorithm SM-2. I didn't notice that when I first published the package on PyPI, and I can't change the package name.

📦  PyPI page

Table of Contents

• Motivation
• Installing and Supported Versions
• A Simple Example
• Features
• What is SM-2?
• Code Reference
• Testing
• Changelog
• Credits

Motivation

The goal was to have an efficient way to calculate the next review date for studying/learning. Removes the burden of remembering the algorithm, equations, and math from the users.

Installation and Supported Versions

Package Install

Install and upate the package using pip:

pip install -U supermemo2

To Play Around with the Code

Download the code:

git clone https://github.com/alankan886/SuperMemo2.git

Install dependencies to run the code:

pip install -r requirements.txt

supermemo2 supports Python 3.8+

A Simple Example

from supermemo2 import first_review, review

# first review
# using quality=4 as an example, read below for what each value from 0 to 5 represents
# review date would default to datetime.utcnow() (UTC timezone) if not provided
first_review = first_review(4, "2024-06-22")
# review prints { "easiness": 2.36, "interval": 1, "repetitions": 1, "review_datetime": "2024-06-23 01:06:02"))

# second review
second_review = review(4, first_review["easiness"], first_review["interval"], first_review["repetitions"], first_review["review_datetime"])
# review prints similar to example above.

Features

📣  Calculates the review date of the task following the SM-2 algorithm.
📣  The first_review method to calculate the review date at ease without having to know the initial values.

What is SM-2?

🎥  If you are curious of what spaced repetition is, check this short video out.

📌  A longer but interactive article on spaced repetition learning.

📎  The SM-2 Algorithm

What are the "values"?

The values are the:

• Quality: The quality of recalling the answer from a scale of 0 to 5.
  • 5: perfect response.
  • 4: correct response after a hesitation.
  • 3: correct response recalled with serious difficulty.
  • 2: incorrect response; where the correct one seemed easy to recall.
  • 1: incorrect response; the correct one remembered.
  • 0: complete blackout.
• Easiness: The easiness factor, a multipler that affects the size of the interval, determine by the quality of the recall.
• Interval: The gap/space between your next review.
• Repetitions: The count of correct response (quality >= 3) you have in a row.

Code Reference

first_review( quality, review_datetime=None**)**

      function that calcualtes the next review datetime for the your first review without having to know the initial values, and returns a dictionary containing the new values.

Parameters:

• quality (int) - the recall quality of the review.
• review_datetime (str or datetime.datetime) - optional parameter, the datetime in ISO format up to seconds in UTC timezone of the review.

Returns: dictionary containing values like quality, easiness, interval, repetitions and review_datetime.

Return Type: Dict

Usage:

from supermemo2 import first_review
# using default datetime.utcnow() if you just reviewed it
first_review(3)

# providing string date in Year-Month-Day format
first_review(3, "2024-06-22")

# providing date object date
from datetime import datetime
d = datetime(2024, 1, 1)
first_review(3, d)

review( quality, easiness, interval, repetitions, review_datetime=None )

      Calcualtes the next review date based on previous values, and returns a dictionary containing the new values.

Parameters:

• quality (int) - the recall quality of the review.
• easiness (float) - the easiness determines the interval.
• interval (int) - the interval between the latest review date and the next review date.
• repetitions (int) - the count of consecutive reviews with quality larger than 2.
• review_datetime (str or datetime.datetime) - optional parameter, the datetime in ISO format up to seconds in UTC timezone of the review.

Returns: dictionary containing values like quality, easiness, interval, repetitions and review_datetime.

Return Type: Dict

Usage:

from supermemo2 import first_review, review
# using previous values from first_review call
r = first_review(3)

# using default datetime.utcnow() if you just reviewed it
review(3, r["easiness"], r["interval"], r["repetitions"])

# providing review_datetime from previous review
review(3, r["easiness"], r["interval"], r["repetitions"], r["review_datetime"])

# providing string review_datetime
review(3, r["easiness"], r["interval"], r["repetitions"], "2024-01-01")

# providing datetime object review_datetime
from datetime import datetime
d = datetime(2024, 1, 1)
review(3, r["easiness"], r["interval"], r["repetitions"], d)

Testing

Assuming you dowloaded the code and installed requirements.

Run the tests

pytest tests/

Check test coverages

pytest --cov

Check coverage on Codecov.

Changelog

3.0.0 (2024-06-22): Major changes/rebuild, Update recommended

• Rewrote the code to remove the class structure, simplfying the code and usability.
• Update to provide datetime instead of just date, more specific with when to review.

2.0.0 (2021-03-28): Major changes/rebuild, Update recommended

• Rebuilt and simplfied the package.

1.0.3 (2021-01-30): Minor bug fix, Update recommended

• Re-evaluate the default date argument to first_review() on each call.

1.0.2 (2021-01-18): Major and Minor bug fix, Update recommended

• Add required attrs package version to setup.py.
• Allow users to access SMTwo model.
• Fix E-Factor calculation when q < 3.

1.0.1 (2021-01-02): Fix tests, update README and add Github actions, Update not required

• Add missing assertions to test_api.py.
• Update README badges and fix format.
• Add Github actions to run tests against Python versions 3.6 to 3.9 in different OS, and upload coverage to Codecov.

1.0.0 (2021-01-01): Complete rebuild, Update recommended

• Build a new SMTwo class using the attrs package.
• Provide API methods to quickly access the SMTwo class.
• Develop 100% coverage integration and unit tests in a TDD manner.
• Write new documentation.

0.1.0 (2020-07-14): Add tests, Update not required

• Add passing unit tests with a coverage of 100%.

0.0.4 (2020-07-10): Minor bug fix, Update recommended

• Fix interval calculation error when q < 3.

0.0.3 (2020-07-06): Documentation Update, Update not required

• Add new section about SM-2 in documentation, and fix some formats in README.

0.0.2 (2020-07-05): Refactor feature, Update recommended

• Refactor the supermemo2 algorithm code into a simpler structure, and remove unnecessary methods in the class.

0.0.1 (2020-07-02): Feature release

• Initial Release

Credits

1. pytest
2. The SM-2 Algorithm