Search Results for: --------

This project involved completing the AWS Academy Cloud Foundations course that had labs and tested what I learned after each module. Primarily, I learned about cloud computing, its advantages, and gained extensive knowledge about the AWS Cloud and the products and services it offers. Through labs, I practiced with Identity and Access Management, Virtual Private Cloud, Elastic Compute Cloud, Elastic Block Store, and Relational Database Service, which are part of the security, networking, compute, storage, and databases service categories. I also learned about cloud architecture and the well-architected framework, scaling, and load balancing. By completing this project, I discovered how the services and resources of cloud technology substitute a traditionally physical system.

I made a decision tree based of a dataset containing the information of patients at a hospital in the ICU to predict mortality rates of the patients. This dataset includes data points like age, gender, BMI, Renal failure, blood pressure, MCHC, Anion Gap, etc. This decision tree shows the combination of all these datapoints and more to predict the mortality rate of the patient, allowing the hospital to make more informed decisions, have a better care system, and keep families and patients more educated on their current situations.

For my class project in MIS 2502 taught by Professor Bauman, I chose to make a decision tree in Python that predicts if a patient has diabetes based on certain health factors. I learned about cleaning data and preparing it for use in a decision tree, as well as how minimum-split and maximum-depth affect how a decision tree is formed.

My project was a decision tree analysis of shootings in the Philadelphia area, and what data is relevant to whether or not a shooting will turn out to be fatal. The data represented the importance of where the individual received the wound. It can be used for agencies such as Philadelphia PD to adjust their incident fatality rate. Data like this is useful to analyze because firearm-related deaths are one of the leading problems in the city and this can be a step in the right direction; overall understanding of crime statistics in the city can lead to the future of non-violence.

For this project I found a new suitable dataset on the internet (drugs200.csv) and applied decision tree analysis to build the prediction of the outcome variable (drug). The data set included about 200 patient with their health information and with the decision tree analysis i was able to create a tree that can predit which drug a patient would need with certain health parameters. The data set had a training set classification accuracy of 95%. My data sheet, decision tree, and answer word document are all submitted on canvas.

-Evangelos

This study and analysis delves into the relationship between employee data and their likelihood to recommend their employer. It utilizes a dataset comprising user IDs linked to various workplace factors like job title, salary, satisfaction level, tenure, and propensity to recommend their company. The main goal is to pinpoint factors that affect employee satisfaction based on a Python decision tree classification algorithm.

A significant part of the study involved optimizing a decision tree model. After testing various combinations of splits and depths, the model’s predictive accuracy is low, with just over 60% correct classification rates on both training and validation sets. This low accuracy, coupled with the potential biases in the data, suggests that the model is not yet suitable for practical business applications. The findings underline the need for more comprehensive data to enhance the model’s reliability and applicability in a corporate environment.

 

My project analyzes factors that correlate to the gender of varies patients with lung cancer. It shows strong correlation in factors such as snoring. I used a decision tree that was used in MIS 2502.

The project was to find a suitable dataset that wasn’t used in class to create and analyze a decision tree. I found a dataset that was a combination of approved and denied loan applicants. From this project, I got more practice with decision trees and analyzing the results to make predictions. I also learned which data can and can’t be used to make a decision tree because I had to search for data that could be split. For the project, I had to identify the best minimum split, the highest probability, the lowest probability, and 4 data points that could be predicted by the tree.

For this project, I was required to find a csv dataset online that had not been used in class before, run a decision tree analysis on the dataset, and answer questions related to the decision tree. I found a dataset on Kaggle that had data on patient glucose level, blood pressure, age, genetic predisposition for diabetes, and other predictors, as well as a 0 or 1 value to show that the patient was diabetic. Using this data, I made a decision tree that showed the probability of a patient being diabetic based on the predictors. After I was done making the decision tree, I answered various questions such as the best minimum split value, which had the highest/lowest probability, and to list some datapoints and their probabilities. This was a solo project, all the work was done myself. I gained a deeper insight into how decision trees work, and I feel more confident in using datasets to make predictions.

 

For my class project, I used data from civil birth registration in New York state. The objective of this class project is to analyze ranked baby names that fall below 50th place and those within or up to the 50th rank. I used a decision tree for my analysis. I found the minimum split was at 20 and maximum to set 3 in order to have more clarity and less branches. I found that a majority of baby names were ranked below 50. They also were not of the ethnicity of asian or non-hispanic. The rarest probability of someone’s baby name being found in the data set were below rank 50, asian ethnicity, and not male. In conclusion, I discovered that the rank of the baby’s name doesn’t significantly impact whether they have a rare name, as both the highest and lowest probabilities were observed among names ranked below the top 50.