Video summary

Tugas Besar, Mata Kuliah Data Science

Main summary

Key takeaways

Educational

Main ideas and lessons conveyed

  • The video explains the final semester (UAS) big assignment for a Data Science course.
  • The assignment is a simple scientific article centered on theoretical analysis and does not require coding or building models.
  • The goal is to train students to think scientifically by:
    • choosing a data science topic,
    • explaining the theory clearly,
    • connecting it to real phenomena/examples,
    • discussing implications (benefits/risks),
    • and ending with conclusions and recommendations.
  • Students must produce original work individually and follow academic writing standards, including structure, citations, references, and formatting.
  • Content must be based on course materials, with options from several topic categories (e.g., statistics, ML, regression, classification, clustering, data ethics).

Detailed instructions / requirements (as presented)

1) Nature of the article (what you must do / not do)

Must be:

  • A scientific article written as theoretical analysis
  • Focused on a data science theme
  • Written individually

Must NOT require (and thus you should not do these):

  • No coding
  • No Python program
  • No machine learning model implementation
  • No dataset use required

2) Core focus and expected thinking process

You must demonstrate understanding of data science theory by:

  • explaining concepts in your own words
  • connecting theory with real examples
  • writing clearly and coherently as scientific writing

The discussion should follow this sequence:

  • choose a science/data topic
  • explain the theoretical basis
  • discuss problems/phenomena related to the topic
  • provide examples of application
  • explain benefits
  • explain risks
  • end with conclusions and recommendations

3) Assignment length and formatting

  • Length: 5 to 7 pages
  • Paper: A4
  • Font: Times New Roman, font size 12
  • Line spacing: 1.5
  • Margins:
    • left: 4 cm
    • right: 3 cm
    • top: 3 cm
    • bottom: 3 cm
  • Text alignment: justified
  • Citation style: APA style (as the bibliography style model)
  • Output: PDF (collected/compiled)

4) Citations and references

  • Use citations where needed; the video indicates using sources accessed in Mendeley.
  • Minimum references: 5
    • At least 2 scientific journals
    • Additional allowed sources:
      • at least 1 book (or lecture material)
      • up to 2 trusted articles from outside sources
  • References should be compiled in PDF format and follow the specified format.
  • Academic integrity rules:
    • Do not copy other articles
    • Do not just rename someone else’s work
    • Each student must have:
      • their own discussion/title
      • their own presentation method
    • Do not use fake references or fake DOIs
    • Do not use sources you haven’t actually checked/read (warning about AI hallucinations)

5) Use of AI tools (GPT/other AI)

AI may be used only as a tool, for example:

  • generating title ideas
  • creating an outline (explicitly recommended)
  • helping understand concepts
  • correcting grammar
  • simplifying theoretical explanations
  • checking logical flow

Prohibited:

  • Do not generate and copy/paste the entire article directly
  • Do not fabricate:
    • data
    • analysis results
    • references
    • DOIs
  • Do not use someone’s personal data in AI without permission
  • Students must verify and edit, because AI can be wrong.

Topic options students can choose from (with required conceptual focus)

A) Basic statistics in data science

Discuss roles/importance of:

  • mean, median, mode
  • standard deviation
  • outliers
  • correlation, etc.

Suitable for more conceptual (often easier) topics.

B) Machine learning overview

Discuss:

  • what machine learning is
  • the relationship between scientific data, AI, and ML
  • everyday use cases

Examples mentioned:

  • product recommendations
  • spam detection
  • graduation prediction
  • customer prediction
  • content recommendation systems

C) Regression (numerical prediction)

Must explain:

  • regression predicts numerical values (e.g., house prices, sales, visitors, exam scores, hotel occupancy)

Required distinction:

  • Regression outputs numbers, unlike classification which outputs categories.

D) Classification (category prediction)

Must explain:

  • classification predicts categories/classes

Examples mentioned:

  • spam vs non-spam email
  • pass vs fail
  • loyal vs disloyal customers
  • credit risk (high/low)
  • high/low risk patients

Should discuss (simple manner):

  • label features
  • training data vs testing data
  • evaluation metrics: accuracy, precision, recall, and AUC (wording in the video was unclear, but evaluation metrics were emphasized)

Important warning:

  • accuracy alone may be insufficient, especially for imbalanced data.

E) Clustering (unsupervised grouping)

Must explain:

  • clustering groups data based on similar characteristics
  • it is unsupervised learning because it uses no labels

Examples mentioned:

  • customer segmentation (loyal / potential / passive)
  • regional grouping
  • student grouping by academic risk
  • product grouping by sales patterns

Required interpretive note:

  • clustering results are not automatically meaningful—humans must interpret clusters.

Example interpretations:

  • Cluster 1 = loyal customers (high frequency + high total spending)
  • Cluster 2 = passive customers (rare shopping + low total value)

F) Data ethics (privacy, security, bias, etc.)

Emphasis:

  • data science is not only about accuracy/models; it also covers:
    • privacy
    • security
    • bias
    • transparency
    • responsibility

What to discuss:

  • use cases involving student data (e.g., attendance systems)
  • facial recognition concerns
  • GPT usage for assignments
  • bias in AI recruitment
  • legality/approval, clear objectives, data minimization, security, transparency, accountability

Must also discuss:

  • benefits and risks (e.g., facial data is sensitive; leakage can be harmful)

Article structure (detailed outline students should follow)

  1. Title

    • Must be specific (avoid overly broad titles like just “data science” or “machine learning”).
    • Should reflect main focus and context (e.g., “Analysis of the Role of Classification in Predicting Student Graduation”).

  2. Author identity

    • Full name and identity including:
      • name
      • study program
      • university
      • email

  3. Abstract

    • Length: ~150 to 250 words (sometimes ~200)
    • Must include:
      • brief background
      • purpose
      • writing method
      • main analysis results
      • brief conclusion

  4. Keywords

    • 3 to 5 keywords related to the article content/theme.

  5. Introduction

    • Must include:
      • background and why the topic matters
      • general development of data science
      • problem/phenomenon
      • reasons for choosing the topic
      • purpose of the article
      • benefits of the article
    • Should not only define concepts; must show the problem and how the article addresses it.

  6. Literature Review

    • Explain main theories that serve as the article’s foundation.
    • Use cited theories from books/journals/articles.
    • Example (for classification predicting graduation):
      • data science
      • machine learning / supervised learning
      • classification features and labels
      • evaluation of classification model

  7. Writing Method

    • Since it is theory-only without coding:
      • use a literature study
      • with a qualitative descriptive approach
    • Should state:
      • compiled from analysis of lecture materials, scientific journals, books, supporting articles, and reliable sources
      • connect concepts with real-life application examples

  8. Results and Discussion (most important section)

    • Must show understanding (not copy-pasted theory).
    • Should include application, benefits, risks, and recommendations.

Topic guidance examples:

  - **Regression**
      - regression predicts numerical values
      - give an example case (e.g., sales prediction)
      - specify input and target variables
      - explain evaluation (e.g., **MAE**, **RMSE**, **R²**)
      - discuss risks of trusting results without considering errors

  - **Classification**
      - predict categories like pass/fail
      - explain possible features (attendance, assignment grades, midterm scores, GPA)
      - define labels (pass/fail)
      - note evaluation beyond accuracy, especially under imbalance

  - **Clustering**
      - grouping without labels (unsupervised)
      - explain the need for human interpretation of clusters

  - **Data ethics**
      - discuss legality/approval, minimization, security, transparency, accountability
      - include benefits and risks (e.g., facial attendance privacy risks)

  1. Conclusion

    • Must answer the article purpose.
    • Should summarize the discussion results.
    • Must include brief suggestions/recommendations.
    • Should be specific and useful (e.g., risk prediction can identify at-risk students, but decisions should not rely solely on the model).

  2. Bibliography / References

    • Use APA style.
    • Minimum 5 references as described above.

Examples of research titles (given as inspiration)

  • Analyze the role of statistics in helping data-based decision making
  • Analyze the importance of outlier detection in data analysis processes
  • Analyze differences between regression and classification in machine learning

Additional examples (paraphrased from video examples):

  • regression application analysis
  • classification for predicting student graduation
  • clustering for customer segmentation
  • overfitting vs underfitting risks
  • importance of splitting training vs testing data
  • bias analysis in machine learning prediction systems
  • ethical issues in student data for academic prediction
  • use of GPT as a data science learning tool
  • data privacy and security challenges in the AI era

Speakers / sources featured

Speaker

  • The video presenter/lecturer (no name provided in the subtitles)

Named/mentioned tools or sources

  • Mendeley (for citation/reference management)
  • GPT chat / ChatGPT (AI tool usage guidance)
  • APA style (citation/bibliography format)
  • Gemini / GPT (mentioned in the context of possible hallucinated references)

Original video