Category: uncategorised

Hands-On Project 2: Classification Algorithms in Google Colab

Objective:

In this assignment, you will work with a classification problem using machine learning techniques, specifically focusing on classification algorithms such as Logistic Regression, K-Nearest Neighbors (KNN), and Support Vector Machines (SVM). You will apply these algorithms to a dataset, evaluate their performance, and analyze the results. This will help you understand the key concepts in Module 3, such as data mining methods, algorithms, and model evaluation.

Learning Outcomes:

• Understand and apply classification algorithms to real-world datasets.
• Evaluate the performance of different classification models.
• Understand the application of machine learning in predictive analytics and its integration in business contexts.
• Gain experience in model evaluation metrics such as accuracy, precision, and recall.

Instructions:

Access .

• Open Google Colab and run the code provided directly from the .
• Open the notebook for Chapter 3: Classification and run the code as provided in the Colab interface.
• Follow the instructions in the notebook and execute each step of the classification process.

Follow the Steps in the Notebook:

• Step 1: Load and explore a classification dataset (e.g., Iris dataset or MNIST dataset).
• Step 2: Preprocess the data (e.g., handling missing values, scaling the data).
• Step 3: Train classification models (Logistic Regression, KNN, SVM).
• Step 4: Evaluate model performance using appropriate metrics such as accuracy, precision, recall, and F1-score.

Ensure you run all the code and take notes on your observations, as this will form the basis of your report.

Write a Report:

After executing the code and analyzing the results, write a 1-2 page report with the following structure:

• Introduction (100 words):
• Briefly describe the dataset you used and the classification task. Summarize the algorithms you will apply.
• Process (300 words):
• Explain the steps taken during the assignment, including data exploration, preprocessing, model training, and model evaluation. Include which classification models you applied and the reasoning behind your choices.
• Findings (200 words):
• Discuss the performance of each classification model. Which model performed best? What did the evaluation metrics (accuracy, precision, recall) indicate about the models?
• Conclusion (100-150 words):
• Conclude by reflecting on the application of classification in real-world scenarios. What key insights did you gain from applying machine learning algorithms to classification tasks?

Deliverables:

1. Report (2-3 pages, 1000-1200 words)- ensure you have a cover page, no specific format is required, you are free to use APA or any professional format.

Bayesian Inference Exercise Enhancing a Medical Diagnosis System

Use Case:

You are part of a healthcare AI startup developing an intelligent diagnostic system. Your goal is to enhance the systems accuracy using Bayesian Inferencean approach that combines prior knowledge with new patient data to predict the likelihood of disease. This system must dynamically update its diagnostic predictions as it encounters new cases.

Youll build a Bayesian model using Python, apply it to a medical dataset (e.g., the UCI Heart Disease dataset), and validate its performance with real-world diagnostic outcomes.

Learning Objectives:

• Understand and apply Bayesian inference in real-world predictive modeling.
• Use domain knowledge to define priors and update beliefs with new evidence.
• Gain practical experience with probabilistic programming using PyMC3 or similar libraries.
• Validate and refine models using statistical and diagnostic performance metrics.

Instructions:

1. Data Collection

• Use the UCI Heart Disease dataset or a comparable public health dataset containing:
• Patient symptoms
• Medical history
• Diagnosed conditions
• Perform preprocessing (clean missing values, encode categorical variables, etc.).

2. Prior Knowledge Integration

• Conduct a brief literature review to determine the prior probabilities of various heart conditions or diseases.
• Cite medical studies or datasets used to determine these priors.
• Clearly explain assumptions and how priors are mathematically incorporated.

3. Bayesian Model Development

• Use PyMC3, PyMC, or TensorFlow Probability to implement your model.
• Your model should:
• Use patient symptoms and history as input features.
• Output the posterior probability of a diagnosis.
• Update dynamically as new data points are introduced.

4. Model Validation

• Reserve a validation/test subset (e.g., 20% of the data).
• Evaluate model performance using metrics such as:
• Predictive accuracy
• Log-likelihood
• Confusion matrix
• ROC/AUC if applicable
• Compare predictions with actual diagnoses.

5. Iteration and Refinement

• Based on validation results:
• Adjust prior distributions
• Modify likelihood functions
• Re-train and re-evaluate the model

6. Final Report

• Submit a well-structured report that includes:
• Introduction to the problem and use case
• Description of dataset and priors
• Explanation of the Bayesian model and assumptions
• Summary of validation and results
• Discussion of findings and potential improvements
• Conclusion and next steps

Submission Requirements

Deliverables:

Python Code File (.ipynb or .py):

• Modular, well-commented code.
• Includes data processing, model development, training, and validation.
• Use markdown or comments to explain each section.

Final Report (.pdf or .docx):

• Include figures such as posterior distributions, ROC curves, and performance tables.
• Use APA citation style.

Sample Data Set: