Machine Learning Tasks

1. Problem Statement: Image Classification by K-Nearest Neighbors (KNN)

Tasks:

Implement a KNN classifier to predict the class labels of a given dataset. You can use any publicly available dataset (e.g., Iris, Wine, Breast Cancer dataset, Animals, etc.).
Write code to train the KNN model and evaluate its performance using appropriate metrics (e.g., accuracy, precision, recall).
Your goal is to classify new, unlabeled images into one of these predefined categories.
K-NN can be used for this task by measuring the similarity (usually using Euclidean distance) between the features of the unlabeled image and the features of labeled images in the dataset.
Choose any dataset of your choice and implement the KNN model.

Tasks:

Build a text classifier using the Naive Bayes algorithm. You can choose a specific domain (e.g., sentiment analysis, spam detection, or topic classification).
Preprocess the text data, train the Naive Bayes model, and evaluate its performance. You can use datasets like the IMDb movie reviews dataset or the 20 Newsgroups dataset.
Choose any dataset of your choice and implement the trained model.

Tasks:

Create a chatbot that can answer user queries based on a specific context (e.g., travel, customer support, or weather).
Develop a chatbot using natural language processing techniques. Train it using relevant text data (e.g., FAQs, customer queries, or Wikipedia articles).
Avoid using pre-trained AI APIs; instead, build your own model.

Tasks:

Submit any of your previously done or new project which you think is your best one to date. It can be anything you did, from a simple "Hello, world!" to a complex plagiarism checking bot to a futuristic autonomous self-driving car.

Make a private git repository and add the following ids to your repository:
- Ganesh Rehan's Github Link
- Raj's Github Link
Inside your repository, make separate folders for each problem statement.
For each problem statement, you can add your working directory including the dataset, code files, and anything else you used.
If possible, add a requirements.txt and README.md file.
Display the result of each problem statement at the end of your file using suitable evaluation metrics like accuracy, precision, recall, F1 score, confusion matrix, Satisfaction Score, Conversation Length, Automation Rate, and anything that you feel is a good metric to the problem statement.
If you couldn't finish all the problem statements, you can still submit the unfinished work as well.
You will be evaluated by the readability of the code and how natural it is.
Any commits past the deadline will not be considered.
There is no deadline for any individual problem statements, and you can work on and submit them anytime you feel like doing so. However, the last date of submission for all the problem statements will be the deadline itself.