FIT 1043 Introduction to Data Science Assignment 2

Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

Assignment Sheet

Unit Name Introduction to Data Science

Unit Code FIT 1043

Assignment Name Assignment 2 (20%)

Aim of this assignment

to conduct predictive analytics, by building predictive models on a dataset using Python in the Jupyter Notebook environment

Learning Outcomes

This assignment assesses the following learning outcomes:

Learning Outcome Number Learning Outcome Description

5   Classify the kinds of data analysis and statistical methods available for a data science project;

6   Locate suitable resources, software and tools for a data science project.

Weighting

This assignment is worth [20%] of your overall grade for this unit.

Requirements

This assignment has the following requirements:

Assignment Type                                                  Individual Task (20%)

Response Format/Hand-in Requirements

There are 2 submissions for this, they are

● Moodle submission

● Kaggle submission (Competition’s Link)

1.) Moodle Submission:

○ Submit the following 2 files (including a Jupyter notebook file (.ipynb) containing your Python code, answers and explanations(if required) to all the questions, and CSV file for your prediction in task A4 respectively)

1. Jupyter notebook file (.ipynb) containing your Python code to all the questions respectively

a. A copy of your working Python code to answer the questions.

b. make use of markdown for any observation explanation/ justification.

2. A csv file of your predictions in task A4

2. Kaggle Submission

The purpose of the Kaggle submission is to provide you with an introductory experience on how machine learning models are evaluated.

Another file, called the “FIT1043-MusicGenre-Submission.csv” consists of data where there are no labels (no ‘music_genre’ column). The whole purpose is to be able to predict those labels for this data set.

You are to output the data to a CSV file that contains 6490 rows (6491 if include the headers) and 2 columns, the column “instance_id” and another column named “music_genre”. A sample file without the ‘music_genre’ entries is also available “99999999-YourName-v1.csv”.

Response Specifications

1.) Moodle Submission Link:

2 separate files (i.e., .ipynb file, and csv file). Zip, rar or any other similar file compression format is not acceptable and will have a penalty of 10%.

2.) Kaggle’s Submission - the csv file with 2 columns (ref. “99999999-YourName-v1.csv”

Due Date

11.55pm (MYT), Tuesday (30 April 2024), Week 9

Disclaimer

Generative AI tools cannot be used for any assessments in this unit.

In this unit, you must not use generative artificial intelligence (AI) to generate any materials or content in relation to your assessment. (see Learn HQ)

Notes:

The main submission must be done via the Moodle site’s submission link.

Kindly refer back to the late penalty on the Assessment tab of Moodle site.

Sanity Checks

● After you are done with the tasks, do sanity checks.

○ Run the code and make sure it can be run without errors.

○ You should never submit code that immediately generates an error (warnings are usually fine) when run!

● Make sure that your submission contains everything we've asked for.

Aim

The main objective of Assignment 2 is to conduct predictive analytics, by building predictive models on a dataset using Python in the Jupyter Notebook environment.

This assignment will test your ability to:

● Read and describe the data using basic statistics,

● Split the dataset into training and testing,

● Conduct multi-class classification using Support Vector Machine (SVM)**,

● Evaluate and compare predictive models,

● Explore different datasets and select a particular dataset that meets certain criteria

● Deal with missing data,

● Conduct clustering using k-means

** Not taught in this unit, you are to explore and elaborate these in your report submission. This will be a mild introduction to life-long learning to learn by yourself.

Data

We will explore the following datasets in Part A (plus a dataset of your choice in Part B):

1. FIT1043-MusicGenre-Dataset.csv

2. FIT1043-MusicGenre-Submission.csv

Format: each file is a single comma separated (CSV) file

Description: These two datasets were derived from a list containing features of the list of songs and their music genre.

Columns: There should be 15 columns consisting of the features of the song and the class/label of the song (Hint: the music_genre column)

Column Header            Description

instance_id an unique ID assigned for each entry

artist_name the name of the artist

track_name the name/title of that song

popularity The popularity of the track. The value will be between 0 and 100, with 100 being the most popular.

acousticness

A confidence measure from 0.0 to 1.0 of whether the track is acoustic. 1.0 represents high confidence the track is acoustic

danceability

Danceability describes how suitable a track is for dancing based on a combination of musical elements including tempo, rhythm stability, beat strength, and overall regularity. A value of 0.0 is least danceable and 1.0 is most danceable.

duration_ms The duration of the track in milliseconds.

energy Energy is a measure from 0.0 to 1.0 and represents a perceptual measure of intensity and activity.

instrumentalness

Predicts whether a track contains no vocals. “Ooh” and “aah” sounds are treated as instrumental in this context. Rap or spoken word tracks are clearly “vocal”. The closer the instrumentalness value is to 1.0, the greater likelihood the track contains no vocal content. Values above 0.5 are intended to represent instrumental tracks, but confidence is higher as the value approaches 1.0.

liveness

Detects the presence of an audience in the recording. Higher liveness values represent an increased probability that the track was performed live. A value above 0.8 provides strong likelihood that the track is live.

loudness

The overall loudness of a track in decibels (dB). Loudness values are averaged across the entire track and are useful for comparing relative loudness of tracks. Loudness is the quality of a sound that is the primary psychological correlate of physical strength (amplitude). Values typically range between -60 and 0 db.

speechinesss

Speechiness detects the presence of spoken words in a track. The more exclusively speech-like the recording (e.g. talk show, audio book, poetry), the closer to 1.0 the attribute value. Values above 0.66 describe tracks that are probably made entirely of spoken words. Values between 0.33 and 0.66 describe tracks that may contain both music and speech, either in sections or layered, including such cases as rap music. Values below 0.33 most likely represent music and other non-speech-like tracks.

tempo

The overall estimated tempo of a track in beats per minute (BPM). In musical terminology, tempo is the speed or pace of a given piece and derives directly from the average beat duration.

valence

A measure from 0.0 to 1.0 describing the musical positiveness conveyed by a track. Tracks with high valence sound more positive (e.g. happy, cheerful, euphoric), while tracks with low valence sound more negative (e.g. sad, depressed, angry).

music_genre

Music genres are represented by the following code:

0 - Alternative

1 - Anime

2 - Blues

3 - Classical

4 - Country

5 - Electronic

6 - Hip-hop

7 - Jazz

8 - Rap

9 - Rock

This data is pre-processed data that was extracted from Spotify and provided on Kaggle. You DO NOT have to download or process/wrangle the data from the original source.

Assignment Tasks:

This assignment is worth 20% of this Unit’s assessment. This assignment has to be done using the Python programming language in the Jupyter Notebook environment. It should also be formatted properly using the Markdown language. Below is an example from a past submission. Note: You need to use Python to complete all tasks.

Example 1

This example has a code cell, the output, which is a rather nice pie chart (with some labels that aren’t ideal) and a short explanation.

Good practice:

As good practice, you should start your assignment by providing the title of the assignment and unit code, your name and student ID, e.g.

Example 2

This is also a sample from past submissions..

Assignment Task(s)                Description

Part A : Classification

A1. Supervised Learning

1. Explain supervised machine learning, the notion of labelled data, and train and test datasets.

2. Read the ‘FIT1043-MusicGenre-Dataset.csv’ file and separate the features and the label

(Hint: the label, in this case, is the ‘music_genre’)

3. Use the sklearn.model_selection.train_test_split function to split your data for training and testing.

A2. Classification (training)

1. Explain the difference(s) between binary and multi-class classification.

2. In preparation for classification, your data should be normalised/scaled.

a. Describe what you understand from this need to normalise data (this is in your Week 7 applied session).

b. Choose and use the appropriate normalisation functions available in sklearn.preprocessing and scale the data appropriately.

3. Use the Support Vector Machine algorithm to build the model.

a. Describe SVM. Again, this is not in your lecture content, you need to do some self-learning.

b. In SVM, there is something called the kernel. Explain what you understand from it.

c. Write the code to build a predictive SVM model using your training dataset.

(Note: You are allowed to engineer or remove features as you deem appropriate)

4. Repeat Task A2.3.c by using another classification algorithm such as Decision Tree or Random Forest algorithms instead of SVM.

A3. Classification (prediction)

1. Using the testing dataset you created in Task A1.3 above, conduct the prediction for the ‘music_genre’ (label) using the two models built by SVM and your other classification algorithm in A2.4.

2. Display the confusion matrices for both models (it should look like a 10x10 matrix). Unlike the lectures, where it is just a 2x2, you are now introduced to a multi-class classification problem setting.

3. Compare the performance of SVM and your other classifier and provide your justification of which one performed better.

A4. Independent evaluation

1. Read the ‘FIT1043-MusicGenre-Submission.csv’ file and use the best model you built earlier to predict the ‘music_genre’ for the songs in this file.

2. Unlike the previous section in which you have a testing dataset where you know the ‘music_genre’ class and will be able to test for the accuracy, in this part, you don’t have a ‘music_genre’ and you have to predict it and submit the predictions along with other required submission files.

a. Output of your predictions should be submitted in a CSV file format. It should contain 2 columns: ‘instance_d’ and ‘music_genre’. It should have a total of 6491 lines (1 header, and 6490 entries).

A5. Kaggle Competition

Submit to the Kaggle Submission site with the 2 columns csv file (Obtained from A4.2.a) ) with the naming as

“StudentID-YourName-VersionNumber.csv”

e.g.: 99999999-SicilyTing-v1.csv

Remark: A sample file has been provided

“99999999-YourName-v1.csv”

*Bonus mark on students that are placed at Top 10% of the leaderboard placement.

Part B : Selection of Dataset, Clustering and Video Preparation

B1. Selection of a Dataset with missing data, Clustering

We have demonstrated a k-means clustering algorithm in week 7. Your task in this part is to find an interesting dataset and apply k-means clustering on it using Python. For instance, Kaggle is a private company which runs data science competitions and provides a list of their publicly available datasets: https://www.kaggle.com/datasets

1. Select a suitable dataset that contains some missing data and at least two numerical features. Please note you cannot use the same data set used in the applied sessions/lectures in this unit. Please include a link to your dataset in your report. You may wish to:

● provide the direct link to the public dataset from the internet, or

● place the data file in your Monash student - google drive and provide its link in the submission.

2. Perform k-means clustering, choosing two numerical features in your dataset, and apply k-means clustering to your data to create k clusters in Python (k>=2)

3. Visualise the data as well as the results of the k-means clustering, and describe your findings about the identified clusters.