Data Analysis and Modelling Assignment
1. Course Learning Outcomes Assessed
This assessment supports the following learning outcomes:
- Explain the foundations, applications, limitations, and future directions of Artificial Intelligence (AI) in Science, Technology, Engineering & Mathematics (STEM).
- Identify, analyse and solve real-world STEM problems using AI approaches, algorithms and applications.
- Communicate accurately and collaborate effectively using a variety of tools and techniques specific to the fields of AI and STEM.
2. Overview of Assessment
The digital age has presented us with a priceless commodity – data, and lots of it. However, having lots of data is meaningless if we don’t know how to make sense of all this data. The power of Artificial Intelligence can be harnessed to extract meaning from voluminous amounts of information. This assignment offers you the opportunity to apply the techniques you learn in class to help you make sense of data.
3. Assessment Task
Music therapy (MT) involves using music to help people reduce stress, lift their mood, and enhance their overall mental health. It′s scientifically proven that music can stimulate the release of ″happy″ hormones like oxytocin, which help us feel good. MT isn′t a one-size-fits-all approach. It includes a wide variety of music genres, tailored to fit the unique needs of different organizations and individuals. This means what works for one person might be different for another, making MT a highly personalized form of therapy.
The MxMH dataset is a research project that aims to uncover connections between people′s music preferences and their self-reported mental health. By analyzing this data, researchers hope to improve how MT is applied and gain deeper insights into how music affects the mind. This could lead to more effective treatments and a better understanding of the psychological impact of music.
In summary, MT is a powerful tool for mental wellness, adaptable to various tastes and backed by scientific evidence. Ongoing research like the MxMH dataset helps to fine-tune its application and reveals more about the human mind′s response to music.
We would like you to analyse this dataset and prepare a report.
Dataset Description
Download the data file here: Assignment 2B Dataset (ARFF file Download ARFF file, CSV file Download CSV file)
Dataset Columns
- Age: Respondent′s age
- Primary streaming service: Respondent′s primary streaming service
- Hours per day: Number of hours the respondent listens to music per day
- While working: Does the respondent listen to music while studying/working?
- Instrumentalist: Does the respondent play an instrument regularly?
- Composer: Does the respondent compose music?
- Fav genre: Respondent′s favourite or top genre
- Exploratory: Does the respondent actively explore new artists/genres?
- Foreign languages: Does the respondent regularly listen to music with lyrics in a language they are not fluent in?
- BPM: Beats per minute of favorite genre
- Frequency [columns]: How frequently the respondent listens to classical music Anxiety/Depression/Insomnia/OCD: Self-reported condition on a scale of 0-10
- Music effects: Does music improve/worsen respondent′s mental health conditions?
To help you manage your time so that you are able to successfully complete this assignment, complete the following activities:
Activities
- Review this assignment, the assessment criteria & rubrics.
- Install Weka on your computer and test your installation to familiarise yourself with the environment.
- Read the file in CSV format and convert to ARFF format to get an understanding of what is in the data.
- Describe how ARFF format is different to a CSV format. Write down a brief statement of what is in the data and what you are investigating. Comment on the data and metadata part of the ARFF format.
- Determine the data type of each attribute.
- Determine if your data has any missing values or outliers. Use an appropriate technique to deal with them and discuss why you chose this method.
- Determine if you need to use standardisation or normalisation methods to scale the numerical attributes. Explain which method you used and discuss why.
- Which variables exhibit associations between anxiety, insomnia and depression, if any?
- Did you observe any insights from the data that can help with strategies to recommend new music genres to users based on their listening habits?
- Write down your responses to the above points and use the images to justify your responses.
- Build a model to predict the music effect.
- Build a model for each of the mental health conditions.
- Do you see any aged-based clusters for various mental health conditions?
- Think about these guiding points when solving the above three modeling questions.
- Which machine learning algorithms would you consider for predicting the target variable, and why?
- How would you split the dataset into training and testing sets? Did you observe any class imbalance?
- What techniques would you use to ensure your model′s robustness?
- Which factors affected the model predictions the most? Can you find that out?
- What metrics would you use to evaluate the performance of your models?
What format is required to submit the report?
single PDF file.