01 Learner statistics engine
The ML-Powered Learner Statistics Engine
Lumimetrica is an ML-powered learner statistics engine designed to help educators understand and improve learning outcomes.
It transforms learning activity into structured learner statistics that reveal progress, growth, performance patterns, engagement, risk, and learning trajectories — across individuals, cohorts, programs, and entire learning systems.
Developed and operated by Graph25 Informatics Limited.
- Growth since week 1
- +21
- Cohort percentile
- 81st
- from 37th
- Standard score
- +0.89σ
- Cohort distribution
- Learner L-0417, week 1 → 20
- Share of cohort below
Illustrative chart of a normal distribution of cohort outcomes, measured in standard deviations (σ) from the cohort mean. In week 1, learner L-0417 sat at −0.33σ, the 37th percentile. By week 20 the learner had moved to +0.89σ, the 81st percentile, with 81% of the cohort below.
From learning activity to learning outcomes
- 01 Learning activity
- 02 Statistical signals
- 03 Learner trajectory
- 04 Intelligence
- 05 Intervention
- 06 Learning outcomes
02 Platform
From Learning Activity to Learning Intelligence
Learning environments continuously generate signals: participation, practice, coursework, results, progression. Taken one at a time they are noise. Structured, they describe how learning is actually unfolding.
Lumimetrica organizes these signals into consistent statistical representations of every learner and every group, so that progress can be measured, compared over time, and understood in context.
How learning activity becomes better learning outcomes
- 01
Learning Activity
The everyday record of learning: participation, practice, coursework, results, and progression through content.
- 02
Structured Learner Data
Activity is organized into consistent, time-ordered learner records with shared definitions across groups and programs.
- 03
Statistical Learning
Curated statistical and machine-learning methods estimate progress, growth, variability, and trajectories from those records.
- 04
Learner Intelligence
The results become interpretable learner statistics: where each learner stands, how they are moving, and what is changing.
- 05
Better Learning Outcomes
Educators use that understanding to decide where attention, support, and intervention can matter most.
03 Learner statistics
Understand Every Learner
A single score compresses a great deal of information into one number. Lumimetrica builds a statistical learner profile instead: a set of related measures that together describe where a learner is, how they got there, and where they appear to be heading.
Statistical learner profile
Learner L-0417 · Weeks 1–20
- Progress
- 72% of expected path
- Growth
- +21 points
- Engagement
- 0.81 index
- Consistency
- High
- Learning trajectory
- Improving
- Risk
- Low
- Learning velocity
- 1.1 points per week
- Outcome probability
- 0.78 ± 0.06
Each dimension is estimated from the learner’s own history and interpreted against relevant peers. Read together, they give a far richer view of progress than any single indicator.
Dimensions of the statistical learner profile
- 01 Progress
- How far a learner has advanced along the expected path for their program.
- 02 Growth
- Change in demonstrated learning over a defined period, separated from where the learner started.
- 03 Engagement
- The depth and regularity of participation in learning activity, beyond attendance or log-ins.
- 04 Consistency
- How stable performance is over time, distinguishing steady progress from volatile results.
- 05 Learning trajectory
- The direction and shape of a learner’s path over time: improving, stable, accelerating, or declining.
- 06 Risk
- Statistical indicators that a learner may be moving away from an expected outcome, surfaced for review.
- 07 Learning velocity
- The rate at which a learner is progressing, compared with their own history and with their peers.
- 08 Outcome probability
- The estimated likelihood of reaching a defined learning outcome, always reported with its uncertainty.
04 Longitudinal view
Learning Trajectories
A single score describes a moment. Learning is a process. Two learners with the same score today can be moving in opposite directions.
Lumimetrica observes how each learner’s outcomes evolve over time and estimates the shape of the path, rather than reading isolated results. Trajectories make improvement visible before it shows in a final result — and decline visible while there is still time to respond.
Five illustrative learner trajectories through the cohort distribution
Outcome scale (0–100) · Weeks 1–20 · Cohort mean with ±1σ and ±2σ bands
- Accelerating
- Improving
- Stable
- Declining
- At risk
- Cohort mean (μ)
- ±1σ
- ±2σ
All five learners begin within 15 points of one another, close to the cohort mean. By week 20 they span the distribution, from the 99th percentile down to below the 1st.
Illustrative chart of five fictional learners on a 0–100 outcome scale, drawn over their cohort’s normal distribution. The cohort mean rises from 52 in week 1 to 62 in week 20, with shaded bands at one and two standard deviations. All five learners begin within 15 points of one another. By week 20 their paths have separated across the distribution: accelerating ends at 84 (99th percentile), improving at 70 (81st), stable at 60 (43rd), declining at 50 (9th) and at risk at 34 (<1st).
Illustrative, fictional data. Not real learner or customer records.
Show the illustrative data as a table
| Week | Accelerating | Improving | Stable | Declining | At risk |
|---|---|---|---|---|---|
| 1 | 48.0 | 52.0 | 60.6 | 63.0 | 57.0 |
| 5 | 49.8 | 55.5 | 60.8 | 60.2 | 55.4 |
| 10 | 55.8 | 60.2 | 58.1 | 57.1 | 50.0 |
| 15 | 67.0 | 65.7 | 60.3 | 52.8 | 43.0 |
| 20 | 84.0 | 70.0 | 60.4 | 50.0 | 34.0 |
| Week 20 percentile | 99th | 81st | 43rd | 9th | <1st |
Five trajectory shapes
- Accelerating
- Gains increasing in pace, often after a change in approach or support.
- Improving
- Steady gains at a consistent rate across the period.
- Stable
- Performance holding at a similar level over time.
- Declining
- A gradual loss of ground relative to the learner’s earlier performance.
- At risk
- A falling path that, if sustained, points away from the expected outcome.
05 Scale
From Individuals to Entire Learning Systems
The same statistical foundations apply at every level of a learning organization. Patterns invisible in a single classroom often become clear across a program or a system — and the reverse is also true.
- L1
Individual learner
Progress, growth, engagement, and trajectory for one learner, interpreted against their own history and their peers.
- L2
Cohort
Distributions, variability, and shared patterns across a class, intake, or group.
- L3
Program
How learners move through a course of study, and where outcomes diverge between cohorts.
- L4
Learning organization
Comparable statistics across programs, departments, and sites within one institution.
- L5
Learning system
System-wide views across many organizations, with consistent definitions and the ability to trace trends back down.
Lumimetrica aggregates learner statistics upward without discarding the detail beneath them. A system-level trend can always be traced back to the cohorts, programs, and individual learners that produce it.
06 Machine learning
Detect Patterns Earlier
Many of the patterns that matter most in learning develop gradually. By the time they appear in end-of-term results, the best moment to respond has often passed.
Lumimetrica combines curated statistical methods with machine-learning models to identify emerging patterns in learner data as they form. Each method is chosen for the question it answers, and each output is reported as an estimate, together with the evidence behind it.
Patterns Lumimetrica is designed to surface
-
Unexpected trajectory changes
A learner’s path departs from its established direction.
-
Emerging risk
Indicators accumulate that a learner may be moving away from an expected outcome.
-
Unusual cohort behavior
A group’s distribution shifts or separates in ways that differ from comparable cohorts.
-
Differences in learning velocity
Learners or groups progress at materially different rates through similar material.
-
Persistent performance gaps
Gaps between groups that hold steady over time instead of closing.
-
Changes in engagement
Participation falls or becomes irregular, often before results change.
07 Action
Intelligence for Intervention
Learner statistics become valuable when they support informed action. Lumimetrica is designed to shorten the distance between noticing a pattern and responding to it — while decisions stay with educators.
From signal to monitored outcome
-
Signal detected
A statistically meaningful change is identified in a learner’s or a group’s data.
-
Pattern understood
The signal is placed in context: its size, its persistence, and the evidence behind it.
-
Learners identified
The learners most affected are surfaced for review, individually and as groups.
-
Intervention considered
Educators decide whether and how to respond, drawing on their knowledge of the learners and the setting.
-
Outcome monitored
Subsequent learning is observed, so the effect of the response can be understood over time.
What Lumimetrica provides
- Statistical signals, reported with their uncertainty
- Context from history, peers, and trend
- Identification of learners and groups for review
- Longitudinal follow-up of outcomes
What educators decide
- Whether a signal warrants action
- What form support should take
- How to work with the learners concerned
- When to continue, adjust, or stop
Lumimetrica informs educators. It does not replace their judgment.
08 Groups
Cohort Intelligence
Averages hide most of what is happening in a group. Two cohorts can share almost the same mean while one is tightly clustered and the other is split in two.
Lumimetrica describes cohorts as distributions — their centre, spread, shape, and clusters — and follows how those distributions move over time. Every learner keeps a position within the group, expressed as a percentile that is easy to read without statistical training.
Outcome distributions for two illustrative cohorts
Illustrative dataOutcome scale (0–100) · P10–P90: Percentiles of Cohort A
Illustrative density chart. Cohort A has a mean of 62.0 and a single peak. Cohort B has a mean of 62.0 but two distinct clusters of learners, one below and one above the mean.
| Statistic | Cohort A | Cohort B |
|---|---|---|
| Mean | 62.0 | 62.0 |
| Median | 62.0 | 62.0 |
| Standard deviation | 9.0 | 12.8 |
| Interquartile range | 12.0 | 21.9 |
| Clusters | 1 | 2 |
Percentile position
Reading a cohort statistically
- Distribution
- The full spread of outcomes in a group, not only its average.
- Percentile position
- Where a learner sits within the group. The 75th percentile means ahead of three quarters of their peers.
- Variability
- How widely outcomes differ within a group, and whether that spread is narrowing or widening.
- Clusters
- Sub-groups of learners with similar profiles, which often call for different kinds of support.
Every learner in Cohort B
n = 120 · Median 62.0 · Interquartile range 51.1–72.9
09 Purpose
Everything begins and ends with learning outcomes.
Lumimetrica does not exist to produce more educational data. Most learning organizations already hold more data than they can use. Its purpose is to make learning outcomes observable, understandable, and actionable.
- 01
Observable
Outcomes are measured consistently over time and at every level, so that change can be seen as it happens rather than only at the end of a term.
- 02
Understandable
Statistics are presented with their context — history, peers, and uncertainty — so that each number carries meaning for the people who act on it.
- 03
Actionable
Insight stays connected to the learners and groups it concerns, so it can inform specific, timely decisions about support.
Because Lumimetrica keeps a longitudinal record, it shows not only where outcomes stand but how they have responded to earlier decisions. Each cycle of teaching, support, and review can then be better informed than the last.
Outcome distributions by term, against the expected outcome
Illustrative dataOutcome scale (0–100)
- Outcome distribution
- On track
10 Architecture
Statistical Infrastructure for Learning
Lumimetrica provides a structured statistical layer between learning activity and organizational decision-making.
Instead of every report, team, or tool computing its own version of progress, Lumimetrica maintains one consistent statistical model of learning — defined once, applied everywhere, and updated continuously as new activity arrives.
-
Learning activity
- Learning environments
- Coursework and results
- Participation records
-
Lumimetrica statistical layer
- 01 Structured learner records
- 02 Statistical and machine-learning models
- 03 Learner statistics
- 04 Outcome intelligence
-
Decision-making
- Educators
- Program leaders
- Institutional leadership
- System administrators
- Consistent definitions
- Progress, growth, and risk mean the same thing in every view, every cohort, and every report.
- Longitudinal by design
- Learner histories are preserved, so change over time is always measurable.
- Multi-level
- Statistics for learners, cohorts, programs, organizations, and systems are computed from the same foundations.
- Built for scale
- Designed for complex learning environments, from a single institution to large learning systems.
11 Trust
Security
Learner data is sensitive. Lumimetrica is designed so that protecting it is part of the system’s structure, not an afterthought.
Organizations evaluating Lumimetrica can request detailed security and data-protection information through the contact section below. See also Enterprise Privacy.
- 01
Data protection
Learner and organizational data is handled with protection in mind at every stage, from collection to deletion.
- 02
Encryption
Encryption is used to protect data as it moves between systems and while it is stored.
- 03
Identity and access management
Access is tied to authenticated identities, so that actions are attributable to known users.
- 04
Role-based access
People see only the learners, groups, and statistics that their role requires.
- 05
Auditability
Access to data and significant changes are recorded, so that activity can be reviewed.
- 06
Infrastructure security
The platform runs on maintained infrastructure, with security updates applied as part of normal operations.
- 07
Data isolation
Each organization’s data is kept logically separate from that of every other organization.
- 08
Operational monitoring
Systems are monitored so that unusual activity and operational issues can be detected and investigated.
- 09
Privacy by design
Data collection is limited to what the statistics require, and privacy is considered when features are designed.
- 10
Secure data handling
Data is processed under defined procedures, with attention to retention and secure disposal.
12 Company
About Lumimetrica
Every learner leaves a statistical story.
Lumimetrica is built on a simple belief: better decisions about learning require better visibility into learning outcomes. Learning organizations make consequential decisions every day — about support, programs, and resources — often with only fragments of the evidence they need.
Lumimetrica is developed and operated by Graph25 Informatics Limited, based in Kampala, Uganda. We build statistical infrastructure for learning: careful, transparent, and designed to keep educators at the centre.
Key facts
- Product
- Lumimetrica
- Category
- ML-powered learner statistics engine
- Primary purpose
- Helping educators understand and improve learning outcomes
- Core capabilities
-
- Learner statistics
- Progress analysis
- Growth analysis
- Learning trajectories
- Cohort intelligence
- Pattern detection
- Risk signals
- Outcome monitoring
- Statistical learning
- Operator
- Graph25 Informatics Limited
- questions@graph25.uk
- Telephone
- +256 756 494 518
- Office
- 3rd Floor, Kanjokya House
Plot 90–92 Kanjokya Street
Kampala, Uganda - Website languages
- English · Français · Português
13 Contact Lumimetrica
Questions? Talk to us.
Whether you are exploring Lumimetrica for your organization or have a question about learner statistics, we would be glad to hear from you.
Email us- questions@graph25.uk
- Telephone
- +256 756 494 518
- Office
- Graph25 Informatics Limited
3rd Floor, Kanjokya House
Plot 90–92 Kanjokya Street
Kampala, Uganda