Claude

Plan: Update BITM 330 for Spring 2027

Status: Awaiting instructor approval Working folder: G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus

1. Reconcile the source documents

Read the Fall Word syllabus, spring-2027-schedule.html, and bitm330-schedule.txt from G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\input. Build a field-by-field change list. Use the TXT for the seven labs and cross-check lecture section 2832, Tuesday/Thursday 4:30-5:50 PM, Lecture Center 7. Record conflicting details instead of choosing silently.

2. Map the calendar before editing the document

Read the Spring 2027 section of the official calendar at https://www.albany.edu/registrar/academic-calendar#spring-2027. Enumerate Tuesday/Thursday meetings from the first eligible date, January 21, through May 4. Exclude March 16 and 18 for Spring Break. Map each existing topic, reading, and deliverable in order. Put any unmatched date, including May 4 if unused, in an instructor-decision list. Keep the final-exam date pending unless separately verified.

3. Apply the approved changes and format the syllabus

Update the term, lecture details, seven-row lab table, and dated topic schedule. Preserve all other policies and each original grading weight. Use consistent headings, restrained typography, visible borders on every table cell, repeated headers, and no emojis. Render the Word file to inspect page breaks and long table rows.

4. Create both requested document formats

Save G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\output\live-BITM330-Spring-2027.docx. Create the native Google Doc in the same Drive output folder. Reopen it and compare the tables and text with the Word version. Record its actual URL, or explicitly report a creation blocker.

5. Verify and write the implementation report

Compare every grade weight with Fall and confirm the 100% total. Match all seven labs to the TXT, check every teaching date and spring-break row, search for stale Fall dates, and confirm the input files are untouched. Save the actual checks and instructor questions to G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\output\live-report.md.

Approval requested

Approve this plan before I generate the revised syllabus. Office hours, an unassigned extra teaching day, and an unverified final-exam time remain instructor decisions.

Gemini

Plan: Update BITM330 Syllabus for Spring 2027

Slug: update-bitm330-syllabus-spring-2027 | Repo: cce-2026 | Date: 2026-10-20 | Status: draft

סיכום

• עדכון סילבוס הקורס BITM 330 מסמסטר סתיו 2026 לאביב 2027 על בסיס נתוני לוח הזמנים והמעבדות.
• החלפת פרטי ההרצאה (קבוצה 2832, ימי שלישי וחמישי 16:30–17:50, Lecture Center 7) וטבלת 7 קבוצות המעבדה בנתוני אביב 2027.
• מיפוי מחדש של תאריכי המפגשים (21 בינואר עד 29 באפריל/4 במאי 2027), החלפת חופשות הסתיו בחופשת אביב, ושמירה מוחלטת על שקלול הציונים (100%).
• יצירת קובץ Word מעודכן (.docx) ותוצר תואם ל-Google Docs בתיקיית היעד output/gemini, בצירוף דו"ח שינויים מפורט.

Steps

1. Extract and inspect syllabus structure and schedule data

• Files:
• G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\input\BITM330-Fall-2026.docx
• G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\input\spring-2027-schedule.html
• G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\input\bitm330-schedule.txt
• Do:
• Read Fall 2026 syllabus structure, header tables, course meeting times, lab sections table, topic schedule, and grade weights.
• Parse lecture info from spring-2027-schedule.html (Class #2832, TTH 4:30 PM–5:50 PM, Lecture Center 7).
• Parse the 7 lab sections from bitm330-schedule.txt (4457, 4458, 4459, 4460, 4461, 4462, 4696 with times and rooms).
• Done when:
• All replacement fields (term, lecture time, room, 7 labs) and existing grading weights (100% total) are cataloged.

2. Build the Spring 2027 course calendar and date mapping

• Files:
• G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\input\BITM330-Fall-2026.docx
• Do:
• Map the semester calendar: semester runs Jan 20 to May 4, 2027; Spring Break March 13–19, 2027 (no class Tue 3/16, Thu 3/18).
• First class: Thursday Jan 21, 2027.
• Map all 26–27 lecture topics to the Tuesday/Thursday schedule, preserving topic order, readings, and deliverables.
• Remove Fall Break and Thanksgiving rows; insert Spring Break row.
• Flag any extra day (such as Tuesday May 4) as "Open day (to be planned)" for instructor review.
• Done when:
• A complete date-by-date schedule table for Spring 2027 is generated with holidays and extra days flagged.

3. Generate updated DOCX and Google Doc artifacts in output/gemini

• Files:
• G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\output\gemini\BITM330-Spring-2027.docx
• G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\output\gemini\BITM330-Spring-2027-Syllabus.md
• G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\output\gemini\BITM330-Spring-2027-Syllabus.html
• Do:
• Create directory output\gemini\ if not present.
• Update course title block: Term to Spring 2027, Class number to 2832, room to Lecture Center 7, meeting time to TTH 04:30 PM - 05:50 PM.
• Replace the 7 lab entries in the lab table with the verified Spring 2027 sections, days, times, and rooms (Massry 119 and Taconic B021).
• Replace class schedule table with the mapped Spring 2027 dates.
• Verify every grading weight remains unchanged.
• Save the resulting document as BITM330-Spring-2027.docx via python-docx.
• Provide Google Doc compatible files directly inside the Drive folder for native Google Docs editing.
• Done when:
• BITM330-Spring-2027.docx is saved under output\gemini\ and opens cleanly with styles and formatting intact.

4. Verify deliverables and write the change audit report

• Files:
• G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\output\gemini\report.md
• Do:
• Verify total grade weights sum to 100%.
• Verify all 7 labs match bitm330-schedule.txt.
• Check that no student or FERPA data is present.
• Write report.md summarizing what changed, the exact file paths saved, and specific items for Nim to confirm (e.g., May 4 extra class, office hours, Zoom link).
• Done when:
• report.md exists in output\gemini\ and confirms all verification checks pass.

Relevant files - G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\input\BITM330-Fall-2026.docx - G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\input\spring-2027-schedule.html - G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\input\bitm330-schedule.txt - G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\LIVE-DEMO\2-syllabus\output\gemini\

Verification 1. Read back BITM330-Spring-2027.docx to verify term, class number, meeting times, lab list, and dates. 2. Confirm grade weights total 100%. 3. Confirm all artifacts are written to output\gemini\ and input files remain untouched.

Decisions - Output files are placed in output\gemini\. - All grading percentages, policies, readings, and course prerequisites remain identical to Fall 2026. - The extra semester day (Tuesday, May 4) will be designated as "Open day (to be planned)" for instructor decision.

Claude

Plan: Quick data analysis of BITM330 Fall 2026 attendance log → HTML report

Context

Nim wants a presentation-ready analysis of G:\My Drive\0-Projects\!-important\data-book\publishing\Cengage\cce-2026\examples\data-analysis\BITM330 Fall2026.csv, output as an HTML file, with everything saved in a new claude\ subfolder next to the CSV. This sits in the CCE 2026 examples/ tree, so it doubles as a demo of an AI agent doing a data-analysis task for academic work.

What the data is (from inspection, read-only)

• 2,436 Google-Form-style responses, 9 columns, no name/ID columns.
• One row per student per class session. 8 real sessions (8/25 → 9/29/2026), roughly 250–320 rows each.
• Columns: Timestamp, Date of Class, Class number, Attendance status, Why missed (free text, 14 rows), Seating (filled only on 9/29, n=283), Participation (n=288, multi-select ;-joined), Class question (free text), Your answer (free text).
• Attendance: 2,408 Present, 14 Partial, 14 Absent (98.9% present).
• Submission hours cluster at 2 PM (1,211) and 4 PM (860), suggesting two sections or a two-part class.
• Data-quality issues worth surfacing: ~60 rows where class number disagrees with the date's dominant class; ~20 rows with impossible years (2003–2007, 2027, 3026); 39 rows whose timestamp date ≠ Date of Class; 4 #NAME? Excel artifacts in answers; the same instructor question typed 4–6 different ways.

Privacy constraint

"Why did you miss class" contains health and family details (surgery, funeral, hospital, biopsy). The report must aggregate only (counts by category such as medical / athletics travel / family), never quote excuse text verbatim. Student free-text answers will also be summarized (counts, lengths, yes/no split), not reproduced at length. No names exist in the file, so no other FERPA exposure.

Deliverables (all in ...\examples\data-analysis\claude\)

1. analyze.py — standard-library-only Python (csv, collections, datetime, json, html). No pandas (not confirmed installed; stdlib is sufficient). Reads the CSV, computes all figures, writes the HTML.
2. BITM330-Fall2026-analysis.html — self-contained report: inline CSS, inline SVG charts, no CDN, so it works offline in a conference room. Light/dark aware, readable at phone width.
3. analysis-plan.md — this plan, copied into the folder so the method is visible next to the output.
4. report.md — detailed written report: every number the HTML shows plus the ones that did not fit on a slide (per-session table, normalized question list with counts, answer categories with counts, absence categories, the full data-quality inventory with row counts), method notes, and caveats. Aggregates only, no verbatim student text.
5. README.md — one-paragraph index of the folder and how to re-run (python analyze.py).

Report structure (per Nim's scoping: these five sections only)

1. Submissions per session (bar chart, 8 real dates): shows the drift from ~310 down to ~283.
2. Attendance status (present / partial / absent): overall split plus a per-session breakdown.
3. Class questions: top instructor questions after normalizing case, punctuation, and whitespace; answers shown as categorized counts only (for example yes / no / other for the yes-no question, and coarse themes for the open questions such as "AI use" or "readings"); no verbatim student text.
4. Absence reasons: categorized counts only (medical / athletics travel / family / other).
5. Data-quality notes: the mismatches listed above, framed as "what an agent catches in 30 seconds."

Footer: source file, generated date, method (stdlib script), one-line privacy statement. Dropped from the earlier draft: headline tiles, submission-hour histogram, seating × participation.

Before writing chart code, load the dataviz skill for palette/form guidance (it triggers on any chart).

Steps

1. Create claude\ folder; write analyze.py.
2. Run it; open the HTML; check all five sections render and numbers match the profiling above (2,436 rows; 2,408 present / 14 partial / 14 absent; 14 absence reasons; 8 real session dates).
3. Confirm no verbatim excuse text or long student answers appear in the HTML (grep for "surgery", "funeral", "hospital", "biopsy").
4. Check the page at ~400 px width for horizontal scroll, and in dark mode.
5. Have analyze.py also emit report.md from the same computed figures, so the HTML and the report can never disagree. Copy this plan to analysis-plan.md. Write README.md.
6. Re-read report.md once for the same privacy check as step 3.

Out of scope / not touched

• Nothing in the cce-2026 git repo changes. The CSV is read-only.
• Optional follow-up Nim may want: save the prompt as a .github/prompts/*.prompt.md file in the repo, per AGENTS.md's "where prompts live" rule. Not done unless asked.

DeepSeek

Analysis plan — BITM 330 form data, quick session analysis

Folder: examples/data-analysis/deepseek/ (Google Drive copy of the CCE 2026 project) Source data: ../BITM330 Fall2026.csv — 2,609 form submissions, 9 columns. Read-only; never modified, never copied into the public repo. Prepared: 2026-09-30. Approved by Nim (v2.1) before execution. Executed by analyze.py.

Purpose

A quick, presentation-ready analysis of the BITM 330 class form data for the CCE 2026 data-analysis demo. Five sections only, in this order:

1. Submissions per session — bar chart over the 8 real (in-person) session dates; shows the drift from first to last session.
2. Attendance status — present / partially present / absent, overall and per session.
3. Class questions — the top instructor questions (normalized, case- and punctuation-insensitive), answer categories as counts only, and the yes/no split for the yes/no question.
4. Absence reasons — categorized counts only.
5. Data quality — "what an agent catches in 30 seconds" checklist with counts.

Outputs: report.html (presentation result) plus session-stats.csv and category-counts.csv (backing counts). analysis-notes.md records what was done after the run.

Definitions and rules

• Unit: one form submission. No unique-student inference; no attendance-rate claims. Every figure is labeled as a share or count of submissions.
• Real session dates: parseable 2026 dates with 50 or more submissions (expected: 8/25, 8/27, 9/1, 9/3, 9/8, 9/15, 9/22, 9/29). Dates with fewer rows are "stray" or "near-empty" and are listed in the data-quality section instead of the chart.
• Near-empty meeting days: inferred from the weekday pattern of the eight sessions; calendar days fitting the pattern between the first and last session are checked, and those with ~no submissions are listed.
• No corrections, no deduplication, no imputation. Identical rows are counted, not merged. Entry errors are surfaced, never fixed.
• Question normalization: lowercase, apostrophes removed, other punctuation becomes spaces, whitespace collapsed. The label shown is the most common original spelling of the cluster. "Did not ask" / "no question" style entries are counted separately.
• Answer categories: Yes / No / Unsure / Other / Blank, by keyword rules (priority: blank, unsure, yes, no). Counts only — answer text is never displayed.
• Absence buckets: Illness / medical, Family / personal, Athletics / travel, Schedule / work, Other; first matching bucket wins. Counts only — reason text is never displayed.
• Privacy: free-text columns (reason, question, answer) never appear as raw text in any output. Questions appear only as the top normalized clusters; the displayed labels are reviewed before release.

Method

Single stdlib-only Python script (analyze.py): reads the CSV as utf-8-sig with stripped header names, computes everything in one pass structure, asserts summation checks, writes the three outputs, and prints an ASCII-only run summary.

Expected shapes (expectations, not results)

• ~2,609 rows; 8 real sessions; first-to-last drift roughly 310 -> 283.
• Session 7 (9/15) unusually low (~151 seen in reconnaissance).
• Three near-empty meeting days (9/10, 9/17, 9/24).
• If the run shows different values, the report states the actual values. Expectations are cross-checks, never targets.

Verification

1. Script exits 0; only this folder is written; source file hash unchanged.
2. Summation checks: date buckets = 2,609; per-session attendance = submissions; question buckets = all rows; answer categories = cluster sizes; absence buckets = reasons provided + no-reason count = absences.
3. Expected-vs-actual recorded (8 sessions; drift; 9/15; near-empty days).
4. Privacy scan of all outputs for distinctive free-text tokens; zero hits.
5. Offline check: no external references in report.html; opens via file://; light and dark themes defined.
6. Repo untouched; nothing committed or mirrored.

Provenance

• v2.1 scope approved by Nim on 2026-09-30 (single HTML report + plan + work report, everything in this folder; supersedes the earlier profile/completeness scope).
• Produced for the CCE 2026 hands-on comparison exercise ("deepseek" folder).