Explore the data. Understand the decisions.
Work through the course with interactive explanations, live charts, random-number experiments, probability simulations, and downloadable Excel formula practice.
Interactive course pathway
Complete sections at your paceThis is an independent study companion using the topics from the supplied Fall 2026 course materials. The app is not an official University of Alberta product.
Explore a dataset
Generate numbers, change individual values, and learn why mean, median and mode can tell different stories.
Dataset generator
Random values are generated in your browserIndividual observations
Edit any number below to immediately recalculate.
Frequency distribution
Counts per valueCalculation walkthrough
Test an outlier
See what happens when one observation changesHow spread out are the numbers?
Two datasets can have the same mean but very different amounts of variation. Explore range, variance and standard deviation.
Deviation from the mean
Variance, step by step
Population or sample?
Variance and standard deviation use a different denominator depending on which one you are analyzing.
Find the position in your data.
Percentiles describe where values fall. Quartiles split ordered data into four sections. A box plot turns those cut points into a visual summary.
Interactive box-and-whisker plot
1.5 × IQR whiskers and flagged outliersPercentile explorer
Percentiles and Q1/Q3 use linear interpolation equivalent to Excel PERCENTILE.INC and QUARTILE.INC. Textbooks may use other quartile conventions.
Ordered data
Reading a box plot
Left edge of the box = Q1 (25th percentile). Centre line = median (Q2). Right edge = Q3 (75th percentile). The box width is the interquartile range (IQR). Whiskers reach the most extreme values within 1.5 × IQR; more extreme observations appear as dots.
Random does not mean guaranteed representative.
See how drawing a smaller sample from a fixed population can produce different averages, even when selection is unbiased.
Population: 70% values centred around 40 and 30% around 80. Sampled without replacement, so every member has an equal selection chance.
Simple random sampling
Each population member has the same chance of selection. It reduces selection bias, but a small random sample can still miss important subgroups by chance.
Stratified sampling
Divide the population into mutually exclusive groups, then randomly sample within each group. This can improve subgroup representation if the strata and allocation are chosen appropriately.
Predict the odds. Then test them.
Compare theoretical probability with experimental outcomes. Explore AND, OR and conditional probability using coin tosses and dice.
Compound events: AND
P(A and B) is the chance that both events occur. For independent events: P(A ∩ B) = P(A) × P(B).
Example: tails on a fair coin AND an even number on a fair die = 1/2 × 3/6 = 25%.
Compound events: OR
P(A or B) includes either event or both: P(A ∪ B) = P(A) + P(B) - P(A ∩ B).
Example: tails OR an even number = 1/2 + 1/2 - 1/4 = 75%.
Try a probability question
In a group of 100 workers, 60 have completed training and 25 of those trained workers work on the night shift. What is P(night shift | completed training)?
Independent events do not affect one another. Mutually exclusive events cannot occur together. These are different concepts. For dependent events, P(A and B) = P(A) × P(B | A).
Practice with your own numbers.
Questions use your generated dataset and course-based scenarios. Submit an answer, then review the worked solution.
Your learning progress
Score is saved on this device only. Work through incorrect answers and retry with a new dataset.
From safety observations to better decisions.
Understand why organizations collect safety data, distinguish leading and lagging indicators, and identify the right data type. Based on Module 1 lecture materials and Activities 2 and 3.
Leading indicators
Activities and conditions that can be monitored to improve future performance: planned inspections, training, completed corrective actions and hazard assessments.
Helps assess whether preventive processes are operating. Activity counts alone do not prove risk has been reduced.
Lagging indicators
Outcomes that have already occurred: injuries, lost workdays, equipment damage, severity rates and total recordable injuries.
Useful for trends and accountability, but they cannot by themselves explain causes.
Activity 2 · Classify the safety indicator
Source-based question bankData types · Identify the measurement
Nominal, ordinal, discrete, continuousUnordered categories, such as incident category or department.
Ordered labels, such as low, medium and high risk.
Counts, such as the number of injuries.
Measured quantities, such as noise exposure or temperature.
Activity 3 · Bright Fabrication scenario
Last year: 27 incidents consisting of 20 first aids and 7 equipment breakdowns. The shop has six production machines. Choose an indicator and propose how the owner can measure it.
The numerical values above are illustrative exercises, not facts about Bright Fabrication. When deciding what to track, also consider work hours, equipment usage, maintenance data and worker input.
Make sense of relationships and uncertainty.
Practice how safety data is collected, how errors influence conclusions, how correlations and regression work, and why confidence intervals matter.
Select the right collection method
Correlation and linear regression
Activity 1: course datasetsCorrelation describes association, not causation. Predictions outside the observed data range are extrapolations and have greater uncertainty.
Confidence interval explorer
A confidence interval expresses the precision of a population-mean estimate when the sampling assumptions are met.
Data quality and common errors
Unpredictable variation around a measurement.
A consistent bias in one direction, such as an incorrectly calibrated sensor.
False positive: detecting an effect that is not there.
False negative: missing a real effect.
Check completeness, consistency, exposure denominators, measurement limitations and alternative explanations.
Turn numbers into a clear story.
Compare chart formats from the course, examine visualization pitfalls, and see what basic forecasting can and cannot establish.
Chart explorer · Activity 1
Choose a chart and see a worked safety exampleIllustrative examples for learning. They do not represent actual workplace performance.
Simple forecasting demonstration
Use a hypothetical monthly hazard reporting trend. A linear model can project a direction but cannot prove future safety outcomes.
Training and testing predictive models
The lecture introduces a typical 70% training / 30% testing split. Training data fits the model; test data evaluates its performance on held-out observations.
More reports do not necessarily mean more incidents. Reporting culture, exposure, process changes and data quality may change the observed trends.
Practice the calculations in Excel.
Download structured Excel workbooks containing the supplied course examples and blank calculation cells. Or export your current random dataset as a new Excel practice worksheet.
Your current random dataset
Updates whenever you generate a new datasetWorkbook includes raw observations, blank answer cells, and Excel formula hints.
Pre-populated course practice workbooks
Each workbook has an exercise sheet, blank response cells, and a formula-help sheet. Use the original lecture and activity PDFs to check your interpretation.
Excel formula guide
Select a formula to copy it. The examples assume your observations start in B2. Adjust the reference to match your worksheet.