Regstats

This tool looks for anomalies in registration bumps or dimps, early drops, late drops, and waitlists to help predict future course needs.

Anomolies are detected by looking at whether a certain phenomenon (like registration) is outside the standard deviation. It also indicates the number of students impacted by the “anomaly”.

It also looks at “squeezes”, where the number of seats at the end of the term is significantly less than what we’d expect from regular attrition.

You can use threshold filters to include or exclude more courses. This includes as using more or less than 1 standard deviation to detect irregularities and changing the minimum number of students affected.

What do the regstats columns mean?

count: total enrollment for the course
pct_sd: percent of students outside the mean affected compared to standard deviation. Computed as (mean – count) / sd
impacted: a rough count of how many students outside the mean are in the bump/dip/etc. This is to filter out courses in which not many students are being affected.

Thresholds

Regstat’s primary function is to identify courses of interest. To do so, it needs thresholds of what constitutes a significant deviation from the norm. Here are the thresholds set in config.R and can be adjusted as need be.

sd_buffer

% of standard deviation before flagging as concern. Default: 1 (meaning use exactly the calculated SD)

min_impacted

min difference b/w enrollment and mean (= number of students affected). Default: 6

min_pct_sd

percent of students outside the mean compared to standard deviation. Z = (mean – count) / sd. Default: 1.2

min_squeeze

squeeze is the ratio of avail seats to drops after census. Default: .5 (determined by general inspection).

min_wait

minimum number of students on a wait list before course is flagged. Default: 20

section_proximity

how close to integer before rounding up/down for recommended sections? closer to .5 reduces -100s: Default: .3 (set closer to .5 for more rounding; set closer to 0 for no rounding.

Command Line

You can create the same regstats reports as with the online tool, but also sets of CSV files for groups of courses.

Use regstats to create a list of courses, specifying a term or two PRIOR to what you will forecast for (to forecast for summer, run regstats or 202510 (and/or 202480):

Rscript cedar.R -f regstats –t 202480,202510

This creates a series of files in the output/csv folder:

bumps.csv: Courses with enrollment bumps
after_bumps.csv: the top 10 courses students take after “bump” courses; useful for deciding whether some might be offered in summer, for instance.
dips.csv: Enrollment dips and the courses students take afterward (we might not need as many)
drops.csv: Courses with an unusually high number of drops, since we might need more of those courses
waits.csv: Courses with long wait lists, where a minimum number of students is defined in config.R
all_flagged_courses: all courses gathered into a master list for convenience

Then, use the forecast function and set the -c flag to all_flagged_courses.csv. For instance, to forecast for summer 2025:

Rscript cedar.R -f forecast -c all_flagged_courses.csv -t 202560

AGAIN: just forecasting for summer won’t tell us about NEW courses that might be needed (but regstats and forecast-report will give you ideas!)

As with all reporting tools, to generate an HTML or ASPX report, use –output [csv/html/aspx]. If you don’t specify a format, the default is html. Files are saved to output/regstats-reports

Specify csv to create the series of CSV files for each category (dips/bumps/drops/etc) in the output/csv folder. Files are saved to output/csv

Rscript cedar.R -f regstats -t 202510 --output aspx

How can I automatically generate course reports for courses identified by regstats?

If you want to generate course reports based on the courses identified by regstats, you can use –f course-report -c forecasts -t TERM or TERM_TYPE. This will generate course reports for all courses listed in the forecast_data table that have forecasts for the specified TERM.