NDAB18000U Data Science (DS)
BSc Programme in Computer Science
BSc Programme in Machine Learning and Data Science
This course covers the components that go into a full data science pipeline, such as the collection, processing and cleaning of data, storing it efficiently in a database, the implementation of efficient and modular models, and the exploration of data through visualisations.
Emphasis will be placed on dealing with data from multiple sources, and on the design of a modular workflow.
Finally, the course will touch upon some of the fundamental challenges in data science, such as the presence of bias, and its potential impact on decision-making.
- Reading of structured text:
- Regular expressions and finite automata
- Grammars and parsing
- Central database concepts such as the relational model, data independence, and transactions
- Entity-relation-modelling (ER modelling) and relational data modelling, including transformations from ER modelling to ER-relational data modelling
- Queries in database query-languages, including relational algebra and SQL
- Theory on database normalisation, including functional dependencies, keys, and relational decomposition
- ACID (atomicity, consistency, isolation, durability) properties and use of transactions
Fundamentals of data integration:
- Strategies for dealing with data heterogeneity
- Data cleaning, error handling, and missing data
- Unstructured to structured data
Model design and implementation:
- Basic modelling concepts
- Structured model design
- Model testing and deployment
Data exploration and visualisation:
- Key principles of visualisation
- Dimensionality reduction techniques
- Fundamental visualisation and interaction techniques for different data types
- Techniques for building and deploying visualisations on the web
- Writing scripts for data collection and preprocessing.
- Using a parser generator to read structured text.
- Setting up database systems supporting heterogeneous sources of data.
- Designing a modular pipeline for the analysis of a concrete problem.
- Creating visualisation on the web.
The student understands the key challenges in designing an effective data science workflow supporting multiple data sources and multiple types of analysis. In particular, the student:
- can use SQL queries to make meaningful queries in databases
- can solve basic data integration tasks
- is able to design and understand modular data science pipelines
- can produce cross-platform, shareable, visualisations for the web
- can clearly and precisely document data analysis workflows, methodology and results
MASD and MAD, or MatIntro and SS
DMA or DMFS (DMFS can be followed simultaneously in block 3)
LinAlgDat (LinAlgDat can be followed simultaneously in block 4)
- Theory exercises
- Project work
Written feedback is provided as comments to assignment solutions.
Continuous feedback is provided during exercise classes, where students can engage in Q&A with teaching assistants.
- 15 ECTS
- Type of assessment
- Written assignmentWritten examination, 2 hours under invigilationThe exam consists of two parts:
1. A group project developed during the course and documented with a report wherein the individual contributions are stated (60%)
2. A final written 2-hours examination (40%)
- Exam registration requirements
1-3 mandatory assignments, marked as passed/failed, must be passed to be qualified for the exam.
- Written aids allowed
For the final written exam only written aids are allowed.
- Marking scale
- 7-point grading scale
- Censorship form
- No external censorship
Several internal examiners
As the ordinary exam.
(Re)submission of (individual) project no later than 3 weeks prior to the reexam and a new written two-hour test under invigilation.
If a student is not qualified for the exam then qualification can be achieved by submitting and getting the equivalent assignments approved no later than three weeks before the re-exam.
If 10 or fewer students participate in the reexam, the written test is replaced by an oral examination with a duration of 30 minutes without preparation.
Criteria for exam assesment
See learning goals.