Data Academy Honors - Badges

Five badges representing core competencies in data science and machine learning for chemical engineering applications.

---

Badge Requirements

Each badge requires:

1. Quiz Average: 75%+ on relevant section quizzes

2. Assignments: Complete all coding assignments in the section

3. Badge Activity: Complete one activity from the options below

---

Badge 1: Data Wrangler 🔍

“Before you can analyze data, you must understand it.”

Champions: Nadia Null, Otto Outlier, Practice Panda

Lectures: 01-NumPy, 02-Pandas Intro, 03-Intermediate Pandas

Core Competencies

• Load and inspect datasets (CSV, Excel, databases)

• Handle missing data appropriately

• Detect and address outliers

• Perform data transformations and aggregations

• Create exploratory visualizations

Key Skills Demonstrated

• pandas DataFrame manipulation

• Missing value strategies (drop, fill, impute)

• GroupBy operations and pivoting

• Basic matplotlib/seaborn plots

• Data quality assessment

Badge Activities (choose one)

Activity	Description	Deliverable
EDA Report	Perform exploratory analysis on a chemical engineering dataset	Jupyter notebook with visualizations and insights
Data Cleaning Challenge	Take a messy dataset and clean it systematically	Before/after comparison with documentation
Missing Data Investigation	Analyze patterns of missingness in real data	Report on missingness mechanisms and chosen strategy

---

Badge 2: Pattern Seeker 🗜️

“Less is more—find what matters.”

Champions: Dee Dimension, Viz Vizzy, Cora Correlation

Lectures: 04-Dimensionality Reduction

Core Competencies

• Understand the curse of dimensionality

• Apply PCA for linear dimensionality reduction

• Use t-SNE/UMAP for visualization

• Interpret reduced representations

• Choose appropriate number of components

Key Skills Demonstrated

• Variance explained analysis

• Loading interpretation

• Visualization of high-dimensional data

• Feature correlation analysis

• Dimension selection strategies

Badge Activities (choose one)

Activity	Description	Deliverable
Spectra Analysis	Apply PCA to spectroscopic data (IR, Raman, NMR)	Notebook explaining chemical meaning of components
Process Visualization	Reduce high-dimensional process data for monitoring	Dashboard showing 2D representation with labels
Feature Study	Compare PCA vs t-SNE on a chemical dataset	Written comparison with visualizations

---

Badge 3: Model Builder 🎯

“Let me draw you a line through that.”

Champions: Reggie Regression, Ridge & Lasso, Val Validation, Barry Bias-Variance

Lectures: 05-Linear Regression, 06-Regularization, 07-Nonlinear Methods

Core Competencies

• Build and evaluate regression models

• Understand bias-variance tradeoff

• Apply regularization (Ridge, Lasso, ElasticNet)

• Use cross-validation properly

• Avoid data leakage

Key Skills Demonstrated

• Feature scaling and preprocessing

• Hyperparameter tuning

• Model evaluation metrics (MSE, MAE, R²)

• Train/validation/test splits

• Pipeline construction

Badge Activities (choose one)

Activity	Description	Deliverable
Property Prediction	Build a model to predict a chemical/physical property	Notebook with model selection justification
Regularization Study	Compare Ridge vs Lasso on multicollinear data	Report with coefficient analysis
Cross-Validation Demo	Demonstrate why CV prevents overfitting	Tutorial notebook with visualizations

---

Badge 4: Ensemble Master 🌲

“Many trees make a forest of wisdom.”

Champions: Forrest Random, Greta Gradient-Boost, Clara Cluster

Lectures: 08-Ensemble Methods, 09-Clustering

Core Competencies

• Build and tune Random Forest models

• Apply gradient boosting (XGBoost, LightGBM)

• Understand bagging vs boosting

• Apply clustering algorithms (k-means, hierarchical, DBSCAN)

• Evaluate unsupervised learning results

Key Skills Demonstrated

• Feature importance analysis

• Hyperparameter optimization for ensembles

• Cluster validation metrics

• Choosing k for k-means

• Interpreting dendrograms

Badge Activities (choose one)

Activity	Description	Deliverable
Ensemble Comparison	Compare Random Forest vs XGBoost on ChemE data	Notebook with performance and interpretability analysis
Molecule Clustering	Cluster chemical compounds by properties	Report with cluster interpretations
Process Segmentation	Use clustering for batch process analysis	Visualization of operating regimes

---

Badge 5: Uncertainty Expert 📏

“A prediction without uncertainty is just a guess.”

Champions: Quinn Quantify, SHAP Shapley, Professor Pipeline

Lectures: 10-Uncertainty Quantification, 11-Model Interpretability

Core Competencies

• Quantify prediction uncertainty

• Build confidence/prediction intervals

• Apply SHAP for model interpretation

• Explain model predictions locally and globally

• Communicate uncertainty to stakeholders

Key Skills Demonstrated

• Gaussian Process regression

• Ensemble uncertainty methods

• SHAP value computation and plots

• Feature importance comparison methods

• Clear uncertainty communication

Badge Activities (choose one)

Activity	Description	Deliverable
Uncertainty Report	Build a model with full UQ for a design decision	Report suitable for engineering decision-making
SHAP Analysis	Interpret a black-box model using SHAP	Notebook with global and local explanations
Risk Communication	Present model predictions with uncertainties	Presentation or document for non-technical audience

---

Badge Summary Table

Badge	Lectures	Key Concepts	Validation
Data Wrangler 🔍	01-03	pandas, missing data, EDA	Can clean and explore any dataset
Pattern Seeker 🗜️	04	PCA, t-SNE, dimensionality	Can reduce and visualize high-D data
Model Builder 🎯	05-07	Regression, regularization, CV	Can build validated predictive models
Ensemble Master 🌲	08-09	RF, XGBoost, clustering	Can apply advanced ML methods
Uncertainty Expert 📏	10-11	UQ, SHAP, interpretability	Can quantify and explain predictions

---

Badge Activity Submission

For each badge activity:

1. Create a Jupyter notebook or report

2. Include clear documentation of your process

3. Explain your reasoning and choices

4. Submit through the course assignment system

Evaluation Criteria:

• Technical correctness (40%)

• Clear documentation (30%)

• Insight and interpretation (20%)

• Code quality (10%)

---

“Every badge represents not just what you know, but what you can do with data.”