---

module_code: COM747 module_title: Data Science and Machine Learning course_work: CW2 Group Work Component - 01 tutor_name: Iftikhar Afridi semester: S3 2023-24 partner: Ulster University

Criteria 1: Quality of Exploratory Data Analysis (20%) [20]

Marks: 14 - 20 (Distinction)

• Comprehensive EDA with clear, well-structured analysis of the dataset.
• Data cleaning and preparation steps are thoroughly explained (e.g., handling missing values, outliers, encoding).
• Includes meaningful visualizations (e.g., histograms, scatter plots, heatmaps) that are visually appealing and easy to interpret.
• Demonstrates deep insight by connecting findings from EDA to the goals of the logistic regression model.
• Uses appropriate statistical summaries and charts to showcase relationships between variables (e.g., correlations).

Marks: 12 - 13.8 (Commendation)

• Good EDA with some meaningful analysis of the dataset.
• Most steps in data cleaning and preparation are explained but lack minor details.
• Includes useful visualizations that are moderately well presented and explained.
• Insights derived from EDA are present but lack depth or comprehensive interpretation.

Marks: 10 - 11.8 (Pass)

• EDA is basic and covers minimal aspects of the dataset.
• Few or unclear data preparation steps; limited explanation of cleaning/processing.
• Visualizations are provided but are of low quality or poorly explained.
• Limited or unclear connection between the EDA and model goals.

Marks: 0 - 9.8 (Fail)

• EDA is absent, extremely poor, or irrelevant.
• No or very few visualizations, and those present are unclear or poorly formatted.
• Little to no explanation of data cleaning, preparation, or insights.

Criteria 2: Quality of Definitions and Descriptions of Logistic Regression and Technical Methods (20%) [20]

Marks: 14 - 20 (Distinction)

• Provides accurate, detailed, and comprehensive definitions of logistic regression.
• Explains the theory behind logistic regression, including its mathematical foundation (e.g., sigmoid function, odds ratio, cost function).
• Clearly outlines advantages, limitations, and real-world applications of logistic regression.
• References additional methods or concepts (e.g., feature scaling, regularization, cross-validation) with excellent clarity and context.

Marks: 12 - 13.8 (Commendation)

• Offers good definitions and explanations of logistic regression.
• Includes an overview of key concepts but may lack advanced insight or clarity in some areas.
• Some mention of advantages, limitations, or applications is provided but not elaborated on.
• Explanations of supporting methods (e.g., feature scaling) are present but lack depth.

Marks: 10 - 11.8 (Pass)

• Definitions and descriptions are basic or somewhat unclear.
• Limited explanation of mathematical concepts or supporting methods.
• Few or no references to advantages, limitations, or applications.

Marks: 0 - 9.8 (Fail)

• Definitions are unclear, incorrect, or missing.
• No explanation of key concepts, advantages, limitations, or supporting methods.
• Technical understanding is absent or very poor.

Criteria 3: Quality of Demonstration and Interpretation of Results (60%) [60]

Marks: 42 - 60 (Distinction)

• Excellent demonstration of model implementation, with clear coding steps in Python/R.
• Code is well-structured, properly commented, and aligns with best practices.
• Provides a clear walkthrough of the dataset splitting process, model training, and testing.
• Includes comprehensive interpretations of evaluation metrics (e.g., accuracy, precision, recall, F1-score, ROC curve, AUC).
• Explains odds ratios and model coefficients in relation to the dataset and problem context.
• Includes additional elements like hyperparameter tuning or model validation.

Marks: 36 - 41.4 (Commendation)

• Good coding demonstration with clear and understandable steps.
• Code is mostly well-commented but may lack minor details or best practices.
• Provides a walkthrough of basic model steps but may not include additional insights (e.g., hyperparameter tuning).
• Interpretations of results are present but may lack depth or clarity for some metrics.

Marks: 30 - 35.4 (Pass)

• Coding demonstration is basic or unclear in parts.
• Limited commenting or explanation of code.
• Steps in the model-building process are missing or poorly explained.
• Interpretation of results is shallow or does not connect results to the problem context.

Marks: 0 - 29.4 (Fail)

• Coding demonstration is missing or very poor.
• Code is poorly written, uncommented, or incomplete.
• No interpretation of results or evaluation metrics.
• Presentation lacks planning, structure, or coherence.