Nbgrader hints

These are practical hints on using nbgrader for grades and assignments. You should also see the separate autograding hints page if you use that.

General

To export grades, nbgrader export is your central point. It will generate a CSV file (using a custom MyCourses exporter), which you can download, check, and upload to MyCourses.

For comprehensive details, see Exporting Grades from Nbgrader.

If students submit assignments/you use autograding

See also

Autograding

  • In each notebook (or at least the assignment zero), in the top, have a STUDENT_NUMBER = xxx which they have to fill in. Asking each student to include the student number in a notebook ensures that you can later write a script to capture it.

Late Submission

In this section, we will discuss how to handle late submissions and collection in nbgrader. You might need to customize some configurations, which you will write in the nbgrader_config.py file if you are using your own nbgrader configuration, or in the etc.jupyter.nbgrader_config.py.append file if you are using the default configuration (refer to the Nbgrader basics to learn more about the configuration). Either way, this file will be referred to as the configuration file.

Collection

Before autograding, the submissions are collected from the students. The default behavior is to collect all submissions, regardless of the due date. This can become a problem if a student submits the assignment both before and after the due date. In this case, the student’s grade will be based on the last submission.

You can change this behavior by setting the c.ExchangeCollect.before_duedate to True in the configuration file. This will collect the students’ last submission before the due date, or their last late submission if there is no submission before due date.

c.ExchangeCollect.before_duedate = True

The late submission policies in the following section will affect the submissions collected after the due date regardless.

Policies and Plugins

The default policy is none, which does nothing (no penalty assigned).

Another policy provided by the default plugin is zero, which assigns a zero grade to late submissions. You can change the policy by setting the c.LateSubmissionPlugin.policy in the configuration file:

c.LateSubmissionPlugin.policy = 'zero'

In addition, you can set your desired late submission policy for your course. There are some pre-defined policies that you can choose from, or send us a ticket describing the policy and we will try to create it for you.

See also

Requesting a course

The policies are defined in the aalto_nbgrader_late.py file (visible under the path /m/jhnas/jupyter/software/pymod/). So far, the available policies are:

  1. SubMarks: The student’s grade is reduced by a fixed number of marks for each hour the assignment is late (penalty_unit=1 would reduce 1 mark per hour from the submission score).

  2. SubSteps: The student’s grade is reduced by a fixed number of marks every few hours the assignment is late.

  3. SubRatio: The student’s grade is reduced by a fixed ratio for every day the assignment is late.

Moreover, you can provide a list of students to exempt from the late submission policy. You can set the list by adding a line to the nbgrader configuration file (see the example below).

Example

To adopt the SubRatio policy, you can use the following lines:

c.AssignLatePenalties.plugin_class = 'aalto_nbgrader_late.SubRatio'
c.SubRatio.penalty_unit = 0.2
c.SubRatio.student_extensions = {"student1": 3600 * 24, "student2": 3600 * 12}

In this example, the SubRatio policy is used with a penalty ratio of 20% per day. The students student1 and student2 are granted 24 hours and 12 hours extension respectively.

Another way to extend the deadline for some students is using the Formgrader interface, which is explained in the next section. We recommend using one approach or the other, as using both at the same time may cause issues.

Grant Extensions

Another way to extend the deadline for some students is to use the “🗓️ Grant Extension” button on the “Manage Submissions” page for each individual.

Note

  • You can grant extensions to autograded submissions only.

  • The students cannot see the extension granted to them. You can inform them yourself.

Testing releasing assignments, without students seeing

Sometimes instructors want to release and collect assignments as a test, while the course is running. To understand how the solution is simpler than “make a new course”, we need to understand what “release” and “collect” do: they just move files around. So, you can just move them to a different place (called the exchange) instead of the one that all students see. Nbgrader docs sure doesn’t do a good job of explaining it, but behind the scenes it’s quite simple, and that simplicity means it’s easy to control if you know what you are up to…

You can equally move your test files around to a test, instructor-only exchange for your own testing (Actually, this isn’t even needed, you can just copy them directly, test, and put back in the submitted/ directory. But some people want more. So, from the jupyter terminal, we have made these extra aliases:

# Release to test exchange (as instructor):
nbgrader-instructor-exchange release_assignment  $assignment_id
# Fetch from test exchange (as instructor, pretending to be a student):
nbgrader-instructor-exchange fetch_assignment  $assignment_id
# Submit to test exchange (as instructor, pretending to be a student):
nbgrader-instructor-exchange submit $assignment_id
# Collect to test exchange (as instructor):
nbgrader-instructor-exchange collect $assignment_id

This copies files to and from /course/test-instructor-exchange/, which you can examine and fully control. If you are doing this, you probably need that control anyway. These terms match the normal nbgrader terminology.

There’s no easy way to make a switch between “live exchange” and “instructor exchange” in the web interface, but because of the power of the command line, we can easily do it anyway.

(use type -a nbgrader-instructor-exchange to see just what it does.)

Known problems

  • The built-in feedback functionality doesn’t work if you modify the submitted notebooks (for example, to make them run). nbgrader upstream limitation. Contact us and we can run a script that will release the feedback to your students.

Course data

If you use the /coursedata directory and want the notebook to be usable outside of JupyterHub too, try this pattern:

import os
if 'AALTO_JUPYTERHUB' in os.environ:
    DATA = '/coursedata'
else:
    DATA = 'put_path_here'

# when loading data, always os.path.join(DATA, 'the_file.py')

This way, the file can be easily modified to load data from somewhere else. Of course, many variations are possible.

Converting usernames to emails

JupyterHub has no access to emails or student numbers. If you do need to link to email addresses, you can do the following. (Note: the format USERNAME@aalto.fi works for MyCourses upload, this process is not usually needed these days anymore.)

  • ssh to kosh.aalto.fi

  • cd to wherever you have exported a csv file with your grades (for example your course directory, cd /m/jhnas/jupyter/course/$course_slug/files/).

  • Run /m/jhnas/jupyter/software/bin/username-to-email.py exported_grades.csv - this will add an email column right after the username column. If the username column is not the zeroth (counting from zero), use the -c $N option to tell it that the usernames are in the Nth column (zero indexed).

  • Save the output somewhere, for example you could redirect it using > to a new filename. A full example:

    /m/jhnas/jupyter/software/bin/username-to-email.py mycourses_export.csv > mycourses_usernames.csv

This script is also available on github.

Our scripts and resources

Some scripts at https://github.com/AaltoSciComp/jupyter-wiki .

We are soon going to revise all of our instructor info which can be useful to you later.