Coding Assignment Rubric

Total Points: 100

Functionality (40 points)

Criteria	Exemplary (36-40)	Proficient (32-35)	Developing (28-31)	Beginning (0-27)
Code Execution	Code runs perfectly without errors. All cells execute in order. Network trains successfully and produces valid output.	Code runs with minor issues. 1-2 cells may need adjustment. Network trains but with warnings.	Code has several errors but demonstrates understanding. Network trains partially or with significant warnings.	Code has major errors. Network fails to train or produces no meaningful output.
Data Preprocessing	All preprocessing steps implemented correctly: scaling, encoding, train-test split with proper parameters.	Most preprocessing steps correct. Minor issues with one component.	Preprocessing attempted but with errors or missing steps.	Preprocessing missing or fundamentally incorrect.
Network Implementation	Network architecture correctly implemented. All layers, activations, and forward pass work properly.	Network mostly correct with minor issues in one component.	Network structure attempted but with significant errors.	Network structure missing or fundamentally flawed.
Training Loop	Complete training loop with forward pass, loss calculation, backpropagation, and weight updates.	Training loop mostly complete but missing one minor component.	Training loop attempted but with logical errors or missing major component.	Training loop missing or non-functional.

Code Quality (25 points)

Criteria	Exemplary (23-25)	Proficient (20-22)	Developing (17-19)	Beginning (0-16)
Documentation	Extensive comments explaining logic. Clear markdown cells describing each section. Variables well-named.	Good comments on complex sections. Most cells documented. Clear variable names.	Minimal comments. Some cells documented. Variable names could be clearer.	Few or no comments. Poor documentation. Unclear variable names.
Code Organization	Logical flow and structure. Functions used appropriately. No redundant code. Follows best practices.	Generally well-organized. Minor redundancy or style issues.	Some organization but could be improved. Redundant sections present.	Poorly organized. Difficult to follow. Significant redundancy.
Error Handling	Proper error checking and handling. Edge cases considered. Informative error messages.	Basic error handling present. Most cases covered.	Minimal error handling. Some cases not considered.	No error handling. Code breaks easily.

Visualization and Analysis (20 points)

Criteria	Exemplary (18-20)	Proficient (16-17)	Developing (14-15)	Beginning (0-13)
Plots and Graphs	Clear, properly labeled plots. Multiple visualizations showing training progress, accuracy, loss. Professional quality.	Good plots with labels. Shows key metrics. Minor presentation issues.	Basic plots present but lacking labels or clarity. Missing some key visualizations.	Few or poor quality plots. Missing important visualizations.
Results Interpretation	Insightful analysis of results. Discusses accuracy, loss trends, overfitting/underfitting. Makes connections to concepts.	Good analysis of main results. Discusses key metrics appropriately.	Basic analysis. Describes results but limited interpretation.	Minimal or no analysis of results.

Experimentation (15 points)

Criteria	Exemplary (14-15)	Proficient (12-13)	Developing (10-11)	Beginning (0-9)
Parameter Testing	Tests multiple hyperparameters systematically. Documents results clearly. Draws meaningful conclusions.	Tests several parameters. Documents most results. Some analysis.	Tests a few parameters. Limited documentation or analysis.	Minimal or no parameter testing.
Creative Exploration	Goes beyond requirements. Tries advanced techniques. Shows curiosity and initiative.	Completes all required experiments. Shows some initiative.	Completes most experiments. Limited exploration.	Minimal experimentation.

Lab Notebook Rubric

Total Points: 50

Scientific Method (20 points)

Criteria	Exemplary (18-20)	Proficient (16-17)	Developing (14-15)	Beginning (0-13)
Hypotheses	Clear, testable hypotheses stated before experiments. Based on understanding of concepts.	Hypotheses stated for most experiments. Generally appropriate.	Some hypotheses present but vague or unclear.	Few or no hypotheses stated.
Methodology	Detailed description of experimental setup. Reproducible procedures. Variables controlled.	Good description of methods. Generally reproducible.	Basic description. Some details missing.	Minimal description. Not reproducible.
Observations	Detailed, objective observations. Quantitative and qualitative data recorded.	Good observations. Most data recorded.	Basic observations. Some data missing.	Minimal observations.

Data Recording (15 points)

Criteria	Exemplary (14-15)	Proficient (12-13)	Developing (10-11)	Beginning (0-8)
Experimental Results	All experiments documented with parameters, outcomes, and metrics.	Most experiments documented. Minor gaps.	Some experiments documented. Significant gaps.	Poor documentation of experiments.
Tables and Charts	Clear tables showing experiment parameters and results. Well-organized.	Good tables present. Generally clear.	Basic tables. Organization could improve.	Tables missing or poorly organized.

Analysis and Reflection (15 points)

Criteria	Exemplary (14-15)	Proficient (12-13)	Developing (10-11)	Beginning (0-8)
Conclusions	Insightful conclusions drawn from data. Connects results to neural network concepts. Discusses implications.	Good conclusions. Makes connections to concepts.	Basic conclusions. Limited connection to concepts.	Weak or missing conclusions.
Reflection Questions	Thoughtful, detailed answers to all reflection questions. Shows deep understanding.	Good answers to most questions. Shows understanding.	Basic answers. Limited depth.	Incomplete or superficial answers.

Conceptual Understanding Quiz Rubric

Total Points: 50

Neural Network Fundamentals (20 points)

Questions may include:

• Explain the biological inspiration for neural networks (5 points)
• Describe the role of weights and biases (5 points)
• Explain activation functions and their purpose (5 points)
• Diagram and label network architecture (5 points)

Points	Criteria
4-5	Complete, accurate explanation with examples. Shows deep understanding.
3	Mostly correct. Minor errors or omissions.
2	Partially correct. Significant gaps in understanding.
0-1	Incorrect or missing. Major misconceptions.

Training Process (15 points)

Questions may include:

• Describe forward propagation (5 points)
• Explain backpropagation and gradient descent (5 points)
• Discuss the role of loss functions (5 points)

Points	Criteria
4-5	Thorough explanation with correct terminology. Can explain to others.
3	Good understanding. Minor terminology issues.
2	Basic understanding. Significant gaps.
0-1	Poor understanding. Major errors.

Practical Application (15 points)

Questions may include:

• How to prevent overfitting (5 points)
• Choosing appropriate architecture (5 points)
• Interpreting training curves (5 points)

Points	Criteria
4-5	Demonstrates practical knowledge. Can apply concepts to new situations.
3	Good practical understanding. Minor application errors.
2	Basic knowledge. Struggles with application.
0-1	Cannot apply concepts.

Experimentation Report Rubric

Total Points: 100

Introduction and Background (15 points)

Criteria	Exemplary (14-15)	Proficient (12-13)	Developing (10-11)	Beginning (0-9)
Problem Statement	Clear, specific problem statement. Provides context and motivation.	Clear problem statement. Some context provided.	Basic problem statement. Limited context.	Unclear or missing problem statement.
Background Research	Demonstrates research into neural networks. References concepts from lesson.	Shows basic research. References some concepts.	Minimal background. Few references.	No background or research shown.

Experimental Design (25 points)

Criteria	Exemplary (23-25)	Proficient (20-22)	Developing (17-19)	Beginning (0-16)
Hypotheses	Clear hypotheses for each experiment. Testable and specific.	Hypotheses stated. Generally testable.	Some hypotheses. Could be more specific.	Vague or missing hypotheses.
Variables	Independent and dependent variables clearly identified. Controls explained.	Variables identified for most experiments.	Some variables identified.	Variables not clearly identified.
Procedure	Detailed procedure. Could be replicated by others.	Clear procedure. Generally reproducible.	Basic procedure. Some gaps.	Unclear procedure.

Results (30 points)

Criteria	Exemplary (27-30)	Proficient (24-26)	Developing (21-23)	Beginning (0-20)
Data Presentation	Professional tables and graphs. All experiments documented. Clear labels and legends.	Good data presentation. Most experiments shown.	Basic presentation. Some experiments missing.	Poor presentation. Significant data missing.
Accuracy of Data	All data accurate and properly recorded. Includes confidence intervals or multiple runs where appropriate.	Data generally accurate. Minor errors.	Some data issues. Inconsistencies present.	Inaccurate or fabricated data.

Analysis and Discussion (20 points)

Criteria	Exemplary (18-20)	Proficient (16-17)	Developing (14-15)	Beginning (0-13)
Interpretation	Insightful analysis. Explains patterns and trends. Discusses why results occurred.	Good analysis. Explains main findings.	Basic analysis. Describes results but limited explanation.	Minimal analysis.
Connection to Theory	Explicitly connects results to neural network theory. Uses correct terminology.	Makes connections to theory. Generally correct terminology.	Limited connections to theory.	No connection to theory.

Conclusion (10 points)

Criteria	Exemplary (9-10)	Proficient (8)	Developing (7)	Beginning (0-6)
Summary	Comprehensive summary of findings. Answers original questions. Discusses implications.	Good summary. Answers main questions.	Basic summary.	Weak or missing conclusion.
Future Directions	Proposes thoughtful extensions or improvements. Shows critical thinking.	Suggests some extensions.	Limited suggestions.	No future directions.

Grading Scale

Overall Course Grade Calculation

Component	Weight	Description
Coding Assignment	40%	Functional neural network implementation
Lab Notebook	20%	Documentation of experiments and learning
Conceptual Quiz	20%	Understanding of key concepts
Experimentation Report	20%	Analysis and communication of results

Letter Grade Scale

Percentage	Letter Grade	Description
93-100%	A	Exceptional understanding and execution
90-92%	A-	Excellent work with minor areas for improvement
87-89%	B+	Very good work, solid understanding
83-86%	B	Good work, demonstrates proficiency
80-82%	B-	Satisfactory work, meets requirements
77-79%	C+	Acceptable work, some gaps in understanding
73-76%	C	Adequate work, significant gaps
70-72%	C-	Minimal acceptable work
67-69%	D+	Below expectations
63-66%	D	Significantly below expectations
60-62%	D-	Barely passing
Below 60%	F	Does not meet minimum requirements

Feedback Guidelines for Teachers

Effective Feedback Should:

1. Be Specific
   Good: "Your network architecture is well-designed with appropriate layer sizes"
   Avoid: "Good job on the network"
2. Be Actionable
   Good: "Try using ReLU activation instead of sigmoid in hidden layers for faster convergence"
   Avoid: "Activation functions could be better"
3. Balance Positive and Constructive - Always note what students did well. Frame improvements as growth opportunities.
4. Connect to Learning Objectives - Reference specific concepts from lesson. Help students see how their work demonstrates understanding.

Common Student Mistakes and Feedback

Mistake	Feedback	Point Deduction
Forgot to normalize data	"Data normalization is crucial for neural network training. See installation guide section on StandardScaler."	-5 points
Used wrong activation	"Consider using ReLU for hidden layers - it typically trains faster than sigmoid."	-3 points
No error handling	"Add try-except blocks to make your code more robust. See code quality rubric."	-5 points
Poor documentation	"Add comments explaining your logic, especially for the training loop. See documentation standards."	-8 points
Missing experiments	"Complete all required hyperparameter experiments to earn full credit."	-10 points

Self-Assessment Checklist for Students

Before submitting, verify:

Code Submission

• All cells run without errors
• Code is well-commented
• Variable names are clear
• Results are clearly displayed
• All required experiments completed

Lab Notebook

• All experiments documented
• Hypotheses stated before experiments
• Results recorded in tables
• Reflection questions answered
• Neat and organized

Quiz (Self-Study)

• Reviewed all key concepts
• Can explain neural networks to others
• Understand training process
• Can interpret results

Report

• Clear introduction
• Detailed methods
• Complete results
• Thoughtful analysis
• Professional presentation

Accommodations and Modifications

For Students with Special Needs

Extended Time:

• Provide 1.5x or 2x time for assessments
• Break coding assignment into smaller checkpoints
• Allow submission in multiple parts

Alternative Formats:

• Oral presentation instead of written report
• Video explanation of code
• Partner programming allowed
• Simplified data sets

Reduced Complexity:

• Use pre-written helper functions
• Focus on concepts over implementation
• Allow use of high-level libraries (Keras)
• Simplified network architectures

For English Language Learners

• Provide glossary of technical terms
• Allow use of translation tools
• Accept code comments in native language
• Peer support for report writing
• Extra time for written assessments

For Advanced Students

Enrichment Options:

• Implement backpropagation from scratch
• Work with larger datasets (CIFAR-10)
• Explore advanced architectures
• Research paper presentation
• Mentor other students

Academic Integrity

Acceptable Collaboration

• Discussing concepts with peers
• Helping debug syntax errors
• Sharing general strategies
• Group study sessions

Academic Dishonesty

• Copying code from classmates
• Using online solutions without understanding
• Submitting AI-generated code as own work
• Fabricating experimental results

Proper Citation

# Based on tutorial from: [URL]
# Modified to work with our dataset

Questions about grading? Contact: info@evolveaiinstitute.com