See where you've been. Glimpse where you're going.
Pick a lens and a metric. The Oracle plots every entity over time and forecasts the next 1โ3 months using a linear trend fit.
Loadingโฆ
๐ Trajectory & Forecast-
How the forecast works. Each entity's history is fit with a simple linear regression and extended forward by the chosen horizon. This is a trend projection, not a machine-learning model - it is most trustworthy for entities with โฅ4 data points and stable cadence. Big program changes, holidays, and data-collection gaps will bend reality away from the dotted line.
Coach Cockpit · Teacher Stats
Teacher Performance Profiles
Select a teacher to view their rubric stats, observation history, and coaching notes.
Averaged Comparison
Coach Cockpit ยท Ask the Data
๐ฌ Have a real conversation with your dataset
No data loaded yet - upload a CSV to start.
Ask like you'd talk to a coach.
Your question gets answered using the rows and filters currently loaded.
Documentation
How AIMS Visualizer works
A premium tour of every feature - load, filter, analyze, save, share, and compare with AI.
01 ยท Overview Tab
The analysis workspace
Load your dataset, apply filters, and reveal insights in seconds.
01
Load CSV
Drop in an EM2 (student performance) or GMW (classroom observation) CSV. The visualizer auto-detects the format and reshapes the data accordingly.
Format detection is automatic - no configuration needed.
GMW wide-format sheets are reshaped row-by-row into analyzable long format.
Qualtrics header/meta rows are stripped automatically.
EM2GMW
โ
02
The Filters
Scope the view to exactly what matters: School, Grade, Assessment Type, Performance Band, and a date range. For GMW, Type becomes Curriculum and Band becomes Module, plus a horizontal strip of observer/district/state/teacher filters.
Every filter updates insight cards and charts together.
SchoolGradeTypeBand๐ Date
03
Apply Filters
Click the copper Apply button to recompute KPIs, insight cards and charts. Use Ctrl/โ + click to select multiple values in a single list. Reset filters clears everything instantly.
Ctrl / โ + click = multi-select.
Shift + click = select a range.
Apply triggers a full rerender - safe to spam.
04
Save to Dashboard
Snapshot the current charts into the Dashboard panel. Give it a name (e.g. "Spring Benchmark - Grade 4 Focus") and it's stored locally so you can revisit or compare later.
Captures all visible charts in one bundle.
Stored in your browser - private, instant.
Open the Dashboard tab to review any snapshot.
05
Export Summary CSV
Download a tidy CSV of the current school ร grade summary - averages, weighted points, and band percentages - ready for Excel, Sheets, or further analysis.
Respects all active filters.
Columns include avg_overall_percentage, points_pct, and per-band percentages.
CSV
โ
06
Quickshare & QS Link
Copy a short link that encodes your exact view - filters, active tab, date window, and GMW extra filters. Recipients open the link and land on the same scene, with charts recomputed from their CSV copy.
Short code backed by Firestore (e.g. ?qs=A7F2K9Q).
Carries the dataset kind so GMW views restore the radar and GMW-specific charts.
AI Summary text (if generated) travels with the link.
A
B
โก qs=A7F2K9Q
07
AI Summary
Turn the automatically-detected insight cards (largest drop, most consistent school, top band shift, largest gain, etc.) into a concise narrative suitable for a memo or stakeholder brief.
Generated on-demand - click Generate summary.
Uses the OpenAI proxy, no local model needed.
Locked to the quickshare state so recipients see the same text.
โฆ
โฆ
โฆ
02 ยท Dashboard Tab
Compare, curate, comment
A second surface for reviewing saved snapshots side by side.
01
Saved Charts ยท left panel
Every chart bundle you save from the Overview tab appears as a card here, sorted by most recent. Names are truncated elegantly; timestamps stay readable.
Copper rail indicates a selectable item.
Active selection uses a plum rail + tinted background.
02
Click on multiple charts
Hold Ctrl/โ and click multiple saved entries to compare them simultaneously. The right panel arranges them in a responsive grid.
Click to view one.
Ctrl/โ + click to add another.
Click again to remove from comparison.
โ
โ
โ
03
Right panel Preview
Selected snapshots render as a live preview of their charts - images when captured, canvases when re-rendered - each labeled with its save name and timestamp.
Responsive grid adapts to your window size.
Cards retain their original color system for comparison fidelity.
04
AI Summary & Comparison
Generate a single summary for one snapshot or an academic-report-style comparison across multiple snapshots - trends, differences, and notable patterns, in plain language.
Single-chart mode: concise insight paragraph.
Multi-chart mode: structured comparison report.
โฆ
03 ยท Ask the Data
Talk to your dataset like a coach
A grounded, coach-flavored conversation surface. Ask in plain English, get back specific, actionable answers pulled directly from the rows currently in scope.
01
Grounded by your filters
Every question is answered against a compact data context block built from your current School, Grade, and date selections. Change a filter and the next turn sees the new slice automatically.
Context refreshes when you hit Apply.
Row count and active filters are visible in the header chip.
Memoized between turns so follow-ups stay fast and cheap.
SchoolGradeDate
โ
02
Coach-tone by design
The system prompt casts the model as a veteran instructional coach. Answers favor specifics, invite action, and suggest next coaching moves where the question calls for it.
No corporate BI voice, no generic hedging.
Teacher names are anonymized by default.
Ends with concrete next moves when the prompt implies action.
๐งโ๐ซ
"
Next move
Next move
03
Suggestion chips & session memory
Start with a one-click prompt (weakest dimension, who to observe next, what changed), or free-type. The last several turns stay in context so follow-ups feel continuous.
Four coach-flavored chip starters above the composer.
Clear button resets the conversation without touching filters.
Cmd/Ctrl + Enter to send; plain Enter keeps multi-line formatting.
Weakest dimensionWho to observeWhat changed
04 ยท The Oracle
Longitudinal trends and forecasts
A generalized over-time view. Pick a lens, a metric, and a horizon. The Oracle plots every entity over time and projects the next one to three months using a simple linear-trend fit.
01
Any lens, any metric
Group by School, Grade, Teacher, Observer, Role, State, District, School Type, Curriculum, Module, or individual Rubric Dimension. Metric options adapt to the dataset (EM2 percentages vs GMW rubric scores).
EM2: Average %, % Proficient+, % Not Yet, Row count.
Each entity's history is fit with a least-squares linear regression. The fit is extended forward as a dashed line in the same color as the history. No hidden ML, no black box: the math is auditable and explained on the page.
Horizons: +1, +2, or +3 buckets (or turn forecast off).
Time buckets: Week, Month, Quarter, or Auto (picks by data span).
Disclaimer card on the page explains when to trust the line.
HistoryForecast
03
Insight cards under the chart
After the chart renders, three cards call out the steepest rise, sharpest drop, and most stable entity by fitted slope, so reviewers see the story without having to read the lines themselves.
Ranked by regression slope per bucket.
Live re-rank whenever filters or controls change.
Row counts shown so small-sample entities are visible.
๐ Rise
๐ Drop
๐ง Stable
05 ยท Teacher Stats
Game-style teacher performance profiles
A character-select screen for teachers. Browse rubric stats as RPG-style attributes, compare multiple teachers, and review coaching notes - all powered by GMW observation data.
01
Character select grid
Every teacher in the GMW dataset appears as a premium card with their initials avatar, school, grade, observation count, and a Level badge (Lv. 10โ30) derived from their overall rubric average.
Mini stat bars at the bottom show an at-a-glance fingerprint of all 8 dimensions.
Cards lift on hover with a copper glow for the selected state.
GMW data auto-loads when you open the page - no upload needed.
KH
Lv.24
MR
Lv.19
JL
Lv.27
02
RPG stat sheet & Power Level
Click a teacher to reveal their full stat sheet: a radar chart of all 8 dimensions, HP/MP-style gradient bars with color-coded scores (orange=Emerging, amber=Developing, blue=Achieving), and a pulsing Power Level orb showing their composite score.
Bars animate in with a smooth width transition.
Power Level orb pulses with a soft plum glow.
Radar chart shows the full dimensional profile at a glance.
2.4
03
Multi-select search & comparison
Search and add teachers as chips in a Facebook Messenger-style input. Select two or more to see an overlaid radar chart comparing their profiles, with a gold dashed line showing the averaged score across selected teachers.
Backspace removes the last chip; click โ on any chip to remove it.
Arrow keys navigate the dropdown; Enter selects.
Combined Power Level shows the group average.
Kim H.Maria R.Joe L.
04
Coaching notes accordion
Every observation comment is organized into 8 collapsible sections, one per rubric dimension. Each note shows the observer name, date, and the score given for that observation - so coaches can trace feedback over time.
The Walkthrough Tool captures classroom data across eight dimensions grouped into four categories. Hover any dimension label in the dashboard - radar axes, bar charts, Teacher Stats stat bars, accordion headers, or character-card mini-bars - to see the full definition in context.
01
Curriculum Use
A teacher's skills in facilitating Great Minds lesson components and pedagogical elements.
Lesson Component Facilitation
Pedagogical Elements
02
Student-Centered Instruction
Emphasis on students' cognitive lift and discourse.
Cognitive Lift
Discourse
?
!
โ
03
Gathering and Using Evidence
How teachers use curriculum-embedded formative and summative assessments and lesson activities to provide focused, actionable feedback to students.
Collecting Evidence
Responding and Feedback
04
Student Engagement
Participation in collaborative and independent lesson activities.
Collaborative Engagement
Independent Engagement
07 ยท Dimension Definitions
All eight dimensions, in detail
Full rubric definitions for each dimension. These same descriptions appear as hover tooltips across the dashboard.
01
Curriculum Use
Lesson Component Facilitation
How effectively the teacher enacts each part of the lesson as designed, with attention to the purpose of each component. Focuses on whether transitions, directions, and facilitation moves ensure that students engage in the intended work of each phase (e.g., sense-making, practice, synthesis), rather than just moving smoothly through the lesson.
Appears as a tooltip on: Overview radar & bar charts, Teacher Stats radar, stat bars, accordion headers, character-card mini-bars.
1โบ2โบ3โบ4
02
Curriculum Use
Pedagogical Elements
The degree to which the teacher uses content-appropriate instructional moves to support how students learn the material. Focuses on whether the teacher applies key pedagogical approaches (e.g., modeling, questioning, scaffolding, use of representations or texts) in ways that deepen understanding and align to the demands of the discipline, rather than relying on generic or procedural teaching strategies.
Appears as a tooltip on: Overview radar & bar charts, Teacher Stats radar, stat bars, accordion headers, character-card mini-bars.
โฒโโ โ
03
Student-Centered Instruction
Cognitive Lift
The level of thinking students are doing. Emphasizes whether students are doing the "heavy lifting" (reasoning, problem-solving, explaining) rather than the teacher doing most of the work.
Appears as a tooltip on: Overview radar & bar charts, Teacher Stats radar, stat bars, accordion headers, character-card mini-bars.
โฆโฆ
04
Student-Centered Instruction
Discourse
The quality and structure of student talk. Focuses on how students engage in meaningful discussion, explain ideas, respond to others, and use academic language.
Appears as a tooltip on: Overview radar & bar charts, Teacher Stats radar, stat bars, accordion headers, character-card mini-bars.
05
Gathering and Using Evidence
Collecting Evidence
How the teacher gathers information about student understanding during the lesson. Includes curriculum-embedded assessment data, observation, and use of student work to monitor progress in real time.
Appears as a tooltip on: Overview radar & bar charts, Teacher Stats radar, stat bars, accordion headers, character-card mini-bars.
โโโ
06
Gathering and Using Evidence
Responding and Feedback
How the teacher uses evidence of student thinking to adjust instruction and provide feedback. Looks at responsiveness, specificity of feedback, and whether it moves student learning forward.
Appears as a tooltip on: Overview radar & bar charts, Teacher Stats radar, stat bars, accordion headers, character-card mini-bars.
TS
07
Student Engagement
Collaborative Engagement
The extent to which students engage meaningfully with peers to build understanding. Focuses on whether collaboration involves shared thinking - such as explaining, questioning, and responding to ideas - rather than simply dividing work or participating superficially.
Appears as a tooltip on: Overview radar & bar charts, Teacher Stats radar, stat bars, accordion headers, character-card mini-bars.
08
Student Engagement
Independent Engagement
The extent to which students are actively and productively engaged in individual work that advances their understanding. Focuses on whether students are thinking, applying, and making sense of content on their own, rather than passively completing tasks or waiting for direction.
Appears as a tooltip on: Overview radar & bar charts, Teacher Stats radar, stat bars, accordion headers, character-card mini-bars.
โฆโฆโฆ
Project history
Every commit on this branch tree
A chronological record of what shipped, organized by branch. Merges are highlighted; in-progress branches are labeled in color.
Loading historyโฆ
AI Summary
Select one or more saved graphs, then generate an AI summary.
For multiple graphs, AI compares patterns in academic-report style.