Featured Course
·Technology & AI·IntermediateApplied Machine Learning for Performance
Fundamentals of Data & ML for Coaches & Educators
Enroll
Paid
One-time course price. Also included with membership
- Lifetime access
- 12 comprehensive modules
- Practical exercises included
- Certificate of completion
Build a trustworthy performance-data workflow—from collecting assessment results to producing and evaluating simple predictive models—without configuring a development environment.
About this course
A project-based course using our in-browser IDE or Google Colab as the primary laboratory. Learn to measure, structure, clean, explore, model, evaluate, and make meaningful decisions using performance…
Curriculum
Course curriculum
12 modules · 127 lessons · 35 hours total
Establish that an assessment result is not automatically usable data. Students learn how measurements become part of an information system involving people, devices, definitions, procedures, and decisions.
01
Lesson 1 — From measurement to decision
Free**How does an assessment result become trustworthy, reusable data?** Use this question throughout the module. Every lesson adds one part of the answer. A jump height, sprint time, bar velocity, or wellness score is not useful merely because a device produced a number. The numbe
02
Lesson 2 — The anatomy of a dataset
Free**How does an assessment result become trustworthy, reusable data?** Use this question throughout the module. Every lesson adds one part of the answer. A jump height, sprint time, bar velocity, or wellness score is not useful merely because a device produced a number. The numbe
03
Lesson 3 — Data types and structures
**How does an assessment result become trustworthy, reusable data?** Use this question throughout the module. Every lesson adds one part of the answer. A jump height, sprint time, bar velocity, or wellness score is not useful merely because a device produced a number. The numbe
04
Lesson 4 — Designing tables that can grow
05
Lesson 5 — Data dictionaries, provenance, and ethical use
**How does an assessment result become trustworthy, reusable data?** Use this question throughout the module. Every lesson adds one part of the answer. A jump height, sprint time, bar velocity, or wellness score is not useful merely because a device produced a number. The numbe
06
Guided practice — Repair the flat sheet
**How does an assessment result become trustworthy, reusable data?** Use this question throughout the module. Every lesson adds one part of the answer. A jump height, sprint time, bar velocity, or wellness score is not useful merely because a device produced a number. The numbe
07
Lab 1 — Build an Athlete Assessment Data Model
Create a Google Sheets workbook with Athletes, Assessments, Results, Metric Dictionary, Import Log, and README tabs.
Teach students to collect reliable data before calculating statistics or building models.
01
Lesson 1 — From a coaching question to a measurable question
**When is a change in performance larger than the uncertainty in the measurement?** Module 1 established how a measurement becomes structured data. Module 2 asks the more difficult question: > Should the measurement be trusted enough to support a decision? Performance testing
02
Lesson 2 — Reliability, validity, sensitivity, and specificity
**When is a change in performance larger than the uncertainty in the measurement?** Module 1 established how a measurement becomes structured data. Module 2 asks the more difficult question: > Should the measurement be trusted enough to support a decision? Performance testing
03
Lesson 3 — Standardization and measurement resolution
**When is a change in performance larger than the uncertainty in the measurement?** Module 1 established how a measurement becomes structured data. Module 2 asks the more difficult question: > Should the measurement be trusted enough to support a decision? Performance testing
04
Lesson 4 — Repeated trials and aggregation
**When is a change in performance larger than the uncertainty in the measurement?** Module 1 established how a measurement becomes structured data. Module 2 asks the more difficult question: > Should the measurement be trusted enough to support a decision? Performance testing
05
Lesson 5 — Change, noise, and decision thresholds
**When is a change in performance larger than the uncertainty in the measurement?** Module 1 established how a measurement becomes structured data. Module 2 asks the more difficult question: > Should the measurement be trusted enough to support a decision? Performance testing
06
Lesson 6 — Missingness, noncompliance, and bias
**When is a change in performance larger than the uncertainty in the measurement?** Module 1 established how a measurement becomes structured data. Module 2 asks the more difficult question: > Should the measurement be trusted enough to support a decision? Performance testing
07
Lesson 7 — From protocol to defensible decision
**When is a change in performance larger than the uncertainty in the measurement?** Module 1 established how a measurement becomes structured data. Module 2 asks the more difficult question: > Should the measurement be trusted enough to support a decision? Performance testing
08
Lab 2 — Design and Audit a Testing Protocol
Create a standardized testing protocol with a reliability dashboard using repeated-trial data.
Introduce ETL as a repeatable process rather than a one-time cleanup exercise.
01
Extract, transform, and load
02
CSV, XLSX, JSON, APIs, and device exports
03
Raw data versus analysis-ready data
04
Preserving an immutable raw-data layer
05
Importing files into Sheets
06
Standardizing dates, names, units, and categories
07
Duplicate detection
08
Missing-data strategies
09
Outlier detection versus outlier deletion
10
Wide-to-long and long-to-wide transformations
11
Provenance, logs, and reproducibility
12
Lab 3 — Clean a Messy IoT Export
Preserve raw files, standardize identifiers, normalize units, and produce an ETL quality report.
Teach students to understand a dataset before attempting prediction.
01
Distributions and descriptive statistics
02
Mean versus median
03
Range, variance, standard deviation, and coefficient of variation
04
Percentiles, quartiles, and z-scores
05
Histograms, box plots, and scatterplots
06
Correlation versus causation
07
Group comparisons
08
Individual response versus group average
09
Time and repeated observations
10
Practical versus statistical importance
11
Lab 4 — Build a Performance Profile
Build athlete summary cards, percentile rankings, cohort comparisons, and defensible findings.
Teach the concepts behind SQL before introducing SQL syntax.
01
Why spreadsheets eventually become difficult to manage
02
Tables, rows, columns, and schemas
03
Primary and foreign keys
04
One-to-one, one-to-many, and many-to-many relationships
05
Filtering, sorting, grouping, and aggregation
06
Joining athlete, session, assessment, and result tables
07
Sheets QUERY() as an introduction to SQL
08
Lookup formulas versus joins
09
Data normalization
10
When to remain in Sheets and when to move to a database
11
Lab 5 — Build a Relational Athlete Dataset
Use separate Athletes, Teams, Sessions, Assessments, and Results tables with QUERY() and lookups.
Build the mathematical intuition students need to evaluate models without turning the course into a proof-based statistics class.
01
Population, sample, parameter, and statistic
02
Random variation and uncertainty
03
Sampling distributions
04
Confidence intervals
05
Null and alternative hypotheses
06
Type I and Type II errors
07
Effect size and practical importance
08
Correlation
09
Linear regression and least squares
10
Residuals and prediction errors
11
Why statistical significance does not guarantee coaching relevance
12
Lab 6 — Model a Performance Relationship
Model a relationship in Sheets with correlation, regression, predictions, residuals, R², MAE, and RMSE.
Explain what changes when a statistical relationship becomes a predictive workflow.
01
What machine learning is—and is not
02
Features, labels, predictions, and model parameters
03
Supervised versus unsupervised learning
04
Regression versus classification
05
Training, validation, and testing data
06
Baseline models
07
Loss and evaluation functions
08
Underfitting and overfitting
09
Generalization
10
Data leakage
11
Class imbalance
12
Model complexity versus usefulness
13
Lab 7 — Train/Test Simulation in Sheets
Randomly assign train/test sets, fit models, compare errors, and produce a model card.
Move from predicting a number to predicting a category or supporting a decision.
01
Binary and multiclass classification
02
Probabilities versus class assignments
03
Decision thresholds
04
Confusion matrices
05
Accuracy, precision, recall, and specificity
06
False positives and false negatives
07
Cost-sensitive decisions
08
Logistic regression intuition
09
Decision trees and interpretable rules
10
Why “high risk” is not a diagnosis
11
Lab 8 — Build a Readiness Classification Model
Build a decision rule and baseline classifier with confusion matrix metrics and threshold analysis.
Introduce common performance applications beyond basic regression and classification.
01
Unsupervised learning
02
Similarity and distance
03
Normalization before comparing athletes
04
K-means clustering intuition
05
Athlete profiling versus athlete labeling
06
Time-series structure
07
Trend, seasonality, cycles, and noise
08
Moving averages
09
Forecast errors
10
Choosing the simplest useful method
11
Lab 9A — Athlete Archetypes
Normalize features and work through a simplified clustering exercise with descriptive cluster names.
12
Lab 9B — Monitor a Training Trend
Build rolling averages, expected ranges, trend lines, forecasts, and error comparisons.
Help students translate their visible Sheets workflow into code while teaching them how to evaluate AI-generated code.
01
Formulas and scripts as instructions
02
Variables and data types
03
Strings, numbers, lists, and dictionaries
04
Conditions
05
Loops
06
Functions
07
Reading CSV files
08
DataFrames and tabular data
09
Translating Sheets logic into Python
10
Reading error messages
11
Prompting an AI coding assistant
12
Testing generated code
13
Protecting athlete data when using AI tools
14
Knowing when generated code is plausible but wrong
15
Lab 10 — Translate a Sheets Pipeline
Translate a prior workflow into Python and compare results with Sheets.
Show what common packages automate after students have manually implemented the underlying ideas.
01
What a package is
02
Estimator-style APIs
03
fit, predict, and transform
04
Preprocessing pipelines
05
Why consistent train/test transformations matter
06
Choosing a method based on the question
07
Interpretable versus complex models
08
Why package availability does not justify model use
09
Reading package documentation
10
Reproducing a model across Sheets and Python
11
Lab 11 — Manual Model Versus Package Model
Compare manual Sheets, Sheets function, and scikit-learn/statsmodels models.
Build an end-to-end Google Sheets application using a new or supplied dataset.
01
Capstone overview and requirements
Fifteen required components from data-collection protocol through coach-facing dashboard and decision memo.
02
Appropriate capstone questions
Examples of defensible performance questions versus unsupported claims.
03
Capstone delivery and assessment
Assessment model weighting labs, documentation, concept checks, model critique, and capstone project.
04
Capstone project
Build a complete performance decision-support system with model card, dashboard, and optional Python translation.
What you'll learn
- Core concepts and theory
- Practical applications
- Real-world case studies
- Hands-on exercises
