Guide for Instructors
This document provides instructors with a brief overview of the basic ideas and capabilities of the Autolab system. It's meant to be read from beginning to end the first time.
Users are either instructors, course assistants, or students. Instructors have full permissions. Course assistants are only allowed to enter grades. Students see only their own work. Each user is uniquely identified by their email address. You can change the permissions for a particular user at any time. Note that some instructors opt to give some or all of their TAs instructor status.
The roster holds the list of users. You can add and remove users one at a time, or in bulk by uploading a CSV file in the general Autolab format:
Semester,email,last_name,first_name,school,major,year,grading_policy,courseNumber,courseLecture,sectionor in the format that is exported by the CMU S3 service:
"Semester","Course","Section","Lecture","Mini","Last Name","First Name","MI","AndrewID","Email","College","Department",...
Attention CMU Instructors:
S3 lists each student twice: once in a lecture roster, which lists the lecture number (e.g., 1, 2,...) in the section field, and once in a section roster, which lists the section letter (e.g., A, B,...) in the section field. Be careful not to import the lecture roster. Instead, export and upload each section individually. Or you can export everything from S3 with a single action, edit out the roster entries for the lecture(s), and then upload a single file to Autolab with all of the sections.
For the bulk upload, you can choose to either:
- add any new students in the roster file to the Autolab roster, or to
- update the Autolab roster by marking students missing from roster files as dropped.
Instructors and course assistants are never marked as dropped. User accounts are never deleted. Students marked as dropped can still see their work, but cannot submit new work and do not appear on the instructor gradebook. Instructors can change the dropped status of a student at any time.
Once a student is added to the roster for a course, then that course becomes visible to the student when they visit the Autolab site. A student can be enrolled in an arbitrary number of Autolab courses.
A lab (or assessment) is broadly defined as a submission set; it is anything that your students make submissions (handins) for. This could be a programming assignment, a typed homework, or even an in-class exam. You can create labs from scratch, or reuse them from previous semesters. See the companion Guide For Lab Authors for info on writing and installing labs.
You can tag each assessment with an arbitrary user-defined category, e.g., "Lab", "Exam", "Homework".
Autograders and Scoreboards
Labs can be autograded or not, at your disrcretion. When a student submits to an autograded lab, Autolab runs an instructor-supplied autograder program that assigns scores to one or more problems associated with the lab. Autograded labs can have an optional scoreboard that shows (anonymized) results in real-time. See the companion Guide For Lab Authors for details on writing autograded labs with scoreboards.
A lab has a start date, due date, end date and grading deadline. The link to a lab becomes visible to students after the start date (it's always visible to instructors). Students can submit until the due date without penalty or consuming grace days. Submission is turned off after the end date. Grades are included in the gradebook's category and course averages only after the grading deadline.
Once an assessment is live (past the start date), students can begin submitting handins, where each handin is a single file, which can be either a text file or an archive file (e.g.,
Penalties and Extensions
You can set penalties for late handins, set hard limits on the number of handins, or set soft limits that penalize excessive handins on a sliding scale. You can also give a student an extension that extends the due dates and end dates for that student.
Autolab provides support for a late handin policy based on grace days. Each student has a semester-long budget of grace days that are automatically applied if they handin after the due date. Each late day consumes one of the budgeted grace days. The Autolab system keeps track of the number of grace days that have been used by each student to date. If students run out of grace days and handin late, then there is a fixed late penalty (possibly zero) that can be set by the instructor.
Each lab contains at least one problem, defined by the instructor, with some point value. Each problem has a name (e.g., "Prob1", "Style") that is unique for the lab (although different labs can have the same problem names).
Once an assessment is live (past the start date), students can begin making submissions (handins), where each submission is a single file.
Grades come in a number of different forms:
- Problem scores: These are scalar values (possibly negative) assigned per problem per submission, either manually by a human grader after the end date, or automatically by an autograder after each submission. Problem scores can also be uploaded (imported) in bulk from a CSV file.
- Assessment raw score: By default, the raw score is the sum of the individual problem scores, before any penalties are applied. You can override the default raw score calculation. See below.
- Assessment total score: The total score is the raw score, plus any late penalties, plus any instructor tweaks.
- Category averages: This is the average for a particular student over all assessments in a specific instructor-defined category such as "Labs, or "Exams". By default the category average is the arithmetic mean of all assessment total scores, but it can be overwridden. See below.
- Course Average: By default, the course average is average of all category averages, but can be overidden. See below.
Submissions can be classified as one of three types: "Normal", "No Grade" or "Excused". A "No Grade" submission will show up in the gradebook as NG and a zero will be used when calculating averages. An "Excused" submission will show up in the gradebook as EXC and will not be used when calculating averages.
Overriding Raw Score Calculations
Autolab computes raw scores for a lab with a Ruby function called
raw_score. The default is the sum of the individual problem scores. But you can change this by providing your own
raw_score function in
<labname>.rb file. For example, to override the raw_score calculation for a lab called
malloclab, you might add the following
raw_score function to
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
# In malloclab/malloclab.rb file def raw_score(score) perfindex = score["Autograded Score"].to_f() heap = score["Heap Checker"].to_f() style = score["Style"].to_f() deduct = score["CorrectnessDeductions"].to_f() perfpoints = perfindex # perfindex below 50 gets autograded score of 0. if perfindex < 50.0 then perfpoints = 0 else perfpoints = perfindex end return perfpoints + heap + style + deduct end
This particular lab has four problems called "Autograded Score", "Heap Checker", "Style", and "CorrectnessDeductions". An "Autograded Score" less than 50 is set to zero when the raw score is calculated.
Note: To make this change live, you must select the "Reload config file" option on the
Overriding Category and Course Averages
The average for a category
foo is calculated by a default Ruby function called
fooAverage, which you can override in the
course.rb file. For example, in our course, we prefer to report the "average" as the total number of normalized points (out of 100) that the student has accrued so far. This helps them understand where they stand in the class, e.g., "Going into the final exam (worth 30 normalized points), I have 60 normalized points, so the only way to get an A is to get 100% on the final." Here's the Ruby function for category "Lab":
1 2 3 4 5 6 7 8 9 10 11
# In course.rb file def LabAverage(user) pts = (user['datalab'].to_f() / 63.0) * 6.0 + (user['bomblab'].to_f() / 70.0) * 5.0 + (user['attacklab'].to_f() / 100.0) * 4.0 + (user['cachelab'].to_f() / 60.0) * 7.0 + (user['tshlab'].to_f() / 110.0) * 8.0 + (user['malloclab'].to_f() / 120.0) * 12.0 + (user['proxylab'].to_f() / 100.0) * 8.0 return pts.to_f().round(2) end
In this case, labs are worth a total of 50/100 normalized points. The assessment called
datalab is graded out of a total of 63 points and is worth 6/50 normalized points.
Here is the Ruby function for category "Exam":
1 2 3 4 5 6
# In course.rb file def ExamAverage(user) pts = ((user['midterm'].to_f()/60.0) * 20.0) + ((user['final'].to_f()/80.0)* 30.0) return pts.to_f().round(2) end
In this case, exams are worth 50/100 normalized points. The assessment called
midterm is graded out of total of 60 points and is worth 20/50 normalized points.
The course average is computed by a default Ruby function called
courseAverage, which can be overridden by the
course.rb file in the course directory. Here is the function for our running example:
1 2 3 4 5
# In course.rb file def courseAverage(user) pts = user['catLab'].to_f() + user['catExam'].to_f() return pts.to_f().round(2) end
In this course, the course average is the sum of the category averages for "Lab" and "Exam".
Note: To make these changes live, you must select "Reload course config file" on the "Manage course" page.
For each lab, students can view all of their submissions, including any source code, and the problem scores, penalties, and total scores associated with those submissions, via the handin history page.
The gradesheet (not to be confused with the gradebook) is the workhorse grading tool. Each assessment has a separate gradesheet with the following features:
- Provides an interface for manually entering problem scores (and problem feedback) for the most recent submmission from each student.
- Provides an interface for viewing and annotating the submitted code.
- Displays the problem scores for the most recent submission for each student, summarizes any late penalties, and computes the total score.
- Provides a link to each student's handin history.
The gradebook comes in two forms. The student gradebook displays the grades for a particular student, including total scores for each assessment, category averages, and the course average. The instructor gradebook is a table that displays the grades for the most recent submission of each student, including assessment total scores, category averages and course average.
For the gradebook calculations, submissions are classified as one of three types: "Normal", "No Grade" or "Excused". A "No Grade" submission will show up in the gradebook as NG and a zero will be used when calculating averages. An "Excused" submission will show up in the gradebook as EXC and will not be used when calculating averages.
Manually assigned grades are by default not released, and therefore not visible to students. You can release grades on an individual basis while grading, or release all available grades in bulk by using the "Release all grades" option. You can also reverse this process using the "Withdraw all grades" option. (The word "withdraw" is perhaps unfortunate. No grades are ever deleted. They are simply withdrawn from the student's view.)