Scribd
Upload
6 views2 pages

Assignment 1

Uploaded by

Rupam Pakhira
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
6 views2 pages

Assignment 1

Uploaded by

Rupam Pakhira
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Data Analytics, Autumn 2024

Assignment 1

Recently, former US President and 2024 Republican presidential candidate Donald


Trump was shot at an outdoor rally in Pennsylvania on a Saturday evening. In response,
the intelligence team conducted a nationwide random sampling of individuals to
identify potential suspects. The collected data includes various attributes such as age,
gender, and occupation, which are detailed in the provided dataframe (Synthetic Data).
The last column of the dataframe indicates the probability of each individual being
either a criminal or innocent, with two possible labels: <=0.5 and >0.5. For classification
purposes, these labels represent distinct classes: a probability greater than 0.5 suggests
a potential criminal, while a probability of 0.5 or less indicates innocence.

Given this dataset, perform the following tasks. You may handle missing values (if any)
according to a scheme of your choice.

A. Perform an Exploratory Data Analysis (EDA) on the dataset. EDA may include
frequency distribution, univariate and multivariate correlation analysis, as well as basic
data visualization (remember the EDA tutorial on 16th August). You are required to
prepare a report (Latex Template) in pdf that includes Exploratory Data Analysis (EDA)
and insights gained from each implementation.

B. Implement Naive Bayes model on the given dataset without relying on any machine
learning libraries (e.g., sklearn). Your task is to code the Naive Bayes algorithm from
scratch to classify individuals as either criminal or innocent. You may use basic packages
such as numpy, pandas, and math. No marks will be given if you use any other custom
classifier or ML libraries.

C. Now, implement Naive Bayes, SVM, Decision Tree, and KNN using the sklearn module
to perform the classification task. Compare the performance of the sklearn Naive Bayes
implementation with your custom Naive Bayes implementation from part B.

D. Finally, implement an ensemble model by combining multiple classifiers, including


your custom Naive Bayes implementation (without sklearn), the sklearn version of SVM,
Decision Tree, and KNN. You must write the ensemble code from scratch, without
relying on any ensemble-related libraries.

Before evaluating the performance, we will first conduct a code plagiarism check. The
performance of each implementation will be assessed using our private test dataset. For
evaluation, we will use accuracy and F1 score metrics. Additionally, we will measure the
running time of each implementation.
For this assignment, each team must submit a single Python file (.py) along with a Colab
notebook link. The notebook should include clearly labelled headings for each cell for
the execution. Afterwards, create a plot to visualize the performance comparison of all
your implementations, including the ensemble model. Add this plot and associated
discussion in the report prepared for part A.

We’ll use MS Teams to accept the submissions. Only one member from each team should
submit the assignment deliverables.

You might also like