FIN559: Learn about ChatGPT and appraise if it is a search engine. In your own words explain how ChatGPT works: Big Data, Cloud Computing and Machine Learning Assignment, SUSS, Singapore

Question 1

Learn about ChatGPT and appraise if it is a search engine. In your own words explain how ChatGPT works. Assess if ChatGPT or chatbots can replace search engines. Try raising the above question to ChatGPT and provide a snapshot of the response.

Question 2

Assess the performance of the algorithm and select the best value of K based on the test performance. You don’t need to optimize using a validation set. Note: In practice, we don’t optimize the algorithm parameters based on the test dataset. We create a validation dataset for tuning the hyperparameters.

Question 3

Understand the dataset by performing exploratory analysis. Use Run4 as the target variable. Design a linear regression model to estimate the target using only the best FOUR (4) attributes from the dataset. Provide the linear regression expression and discuss your results and estimate the relevant metrics and values.

Question 4

Perform exploratory data analysis. First, use only the numerical attributes in the dataset and construct an optimal decision tree. Following this, include all the attributes and construct a new decision tree. Note that you need to transform the text attributes into numbers. Compare the performance of both decision trees and select the best one. You are required to specify the parameters of the decision trees and also the performance metrics.

Question 5

You can use the example codes to download and load the data properly into the programming environment. You are required to provide comments in the code to explain the function of the code used for loading the dataset. Perform exploratory data analysis and show a random sample of four (4) images each for the cat and the rabbit.

Design a CNN with two (2) convolutional layers and three (3) dense layers. Employ ‘ReLU’ for activation and MaxPooling. Keep 15% of the training dataset for validation and use at least 10 epochs. Note you have you the validation dataset in the download as test data and have to use validation_split on the training dataset to create the data for validation. Rate the performance of the algorithm.