In Semester I-2018/2019 the project is conducted as a mini hackathon with the collaboration of Malaysia Airlines Berhad. A dataset on flights is released where the students formed groups to build various solutions by applying their knowledge through this course.

Instructions given to the students

1. Form a team of max 3 people. Submission should be as an e-portfolio in a blog (I recommend google site). Please provide the name of the team members, matric number, picture of each of your team member, email, phone number and assigned tasks.

2. The project consists of 3 parts:

a. at least 2 types of data mining solutions: focus for the project is the development and comparison of machine learning models’ performance

b. at least 2 types of data mining solutions development for digital flight analysis task using Malaysia Airlines Berhad dataset. The examples of task that you could accomplish are

i. flight delay prediction (binary classification)

ii. prediction of flight duration (using regression method)

iii. comparison of models for different routes (investigate the difference per route)

iv. classification of flight delay codes

3. In your blog you need to come out with

a. Introduction: Explain how data mining can benefit digital flight business (support your explanation with other commercial news and academic papers). You need to also provide your elevator pitch here.

b. Data exploration: Describe about the dataset features and data distribution. Illustrate your decision making on the model preparation through data exploration technique. Visualize the data and explain your observation.

c. Data preprocessing: Explain the steps for cleaning, select, integrate and transform the dataset. Provide video or snapshot of the preprocessing, before and after.

d. Data analysis: Explain your observation based on descriptive analytic (eg association rules, clustering). Explain your dataset preparation for the model development. Show the original dataset and the dataset after cleaning/ready for model development.

e. Development and comparison of machine learning models’ performance: Develop at least 3 algorithms and compare the performance of the features by using the algorithms. You may either compare the combination of descriptive models and predictive models (eg clustering-based classification enhancement vs standard classification) or just prediction models (eg decision tree vs logistic regression)

f. Results analysis and conclusion: Explain your performance metric. Prepare a table to show the comparison of the performance.

In each part, provide clear explanation of your experiment setting, approach, observation and

codes. You could also provide a video to show how you have developed your models.

4. Prepare a short idea proposal using Power Point and submit by 9^th December 2018 at 11.59 pm. The idea proposal can be later converted into elevator pitch to introduce about your group’s works. The earlier you submit, and it gets approved, the sooner you can begin your project.

Deadline: Submission as a blog before the end of week 14. Penalty will be given if the students submit later than then deadline as follows:

Late 1 day: minus 10% marks

Late 2 days: minus 20% marks

Late 3 days: minus 30% marks

Later than late 3 days: minus 100% marks

The videos of the project presentation can be viewed here