ETL

https://github.com/melissaEYE/Extract-Transform-Load

The objective is to describe Extract, Transform, Load (ETL) process with ERD diagram.

R

Cluster Analysis on Eye Tracking Data

https://github.com/melissaEYE/Cluster-Analysis

The objective of the analysis was to explore the variables that group together (cluster) to find similarities and relationships within the dataset. More specifically, using K-Means cluster analysis, an unsupervised learning method, to sort items into groups by similarity using eye tracking data and age.

Logistic Regression & Naive Bayesian Classification on AIDS Data

https://github.com/melissaEYE/Logistic-Regression-Naive-Bayesian-Classification

The objective of the analysis was to explore the factors that influence CD4 cell levels in the hope that doctors may to prescribe more effective treatments at different stages of the AIDS disease and perhaps better understand the disease trajectory and health care options. More specifically, using a binary logistic regression, was used to estimate the probability of an outcome of CD4 cells above and below a threshold of 50 cells per cubic millimeter (cells/mm3). Given the results of the logistical regression, a second analytical model, a Naive Bayesian Classification was used to estimate the class probability and specifically to determine if this model could improve, accuracy, sensitivity, and specificity.

Conditional Inference Decision Tree (ctree)

https://github.com/melissaEYE/Conditional-Inference-Tree

The objective of the analysis was to explore the factors that impact a woman’s choice of contraception method. The methods include no use of contraception, long-term methods, or short-term methods. More specifically, using a conditional inference tree (ctree) method of supervised learning, the objective is to categorize the choice of contraception method using input variables of demographic and socio-economic characteristics. Understanding what factors affect a woman’s contraceptive choice would be helpful information for the National Family Planning Program as they work to achieve the goals set out by the National Health System in Indonesia.

Neural Networks and Diabetes

https://github.com/melissaEYE/R---Neural-Networks

The objective of the analysis was to explore the factors that impact a positive or negative diagnosis of diabetes. More specifically, using the Neural Networks (NN) method of supervised learning, the objective is to classify the diabetes diagnosis using various input variables to include health, demographics, and family history. Understanding what factors affect the Pima Indian female’s high incidence rate of diabetes may provide some insight into strategies for mitigation.

Association Rules and Lung Cancer

https://github.com/melissaEYE/Association_Rules

The objective of the analysis was to determine co-morbid symptoms for patients who had lung cancer and were required to undergo a lung resection. Understanding which symptoms to be co-morbid can help to improve patient care and medicine prescription. The type of analysis is association rule learning, whereby associations, or sets of frequent items, between variables are explored to try and find insights and hidden relationships in large datasets (Han, Kamber, Pei, 2011). One example may be, that a person who smokes (lhs, antecedent), is frequently coupled with a symptom of cough and weakness (rhs, consequent).

Text Classification (NLP) and Hotel Reviews

https://github.com/melissaEYE/Text-Classification

The objective of the analysis was to determine the sentiment (positive or negative) of patrons who stayed at the hotels. The type of analysis is text classification. The goal of text classification is to extract features within the unstructured patron reviews that represent the hotels, concepts such as "I like" or "I don't like...."

SAS Enterprise Miner

Bagging-Boosting-Random-Forest

https://github.com/melissaEYE/Bagging-Boosting-Random-Forest

The objective of the analysis was identify bad car buys for a second hand car sales business in order to help them avoid cars that were bad and develop an inventory of quality cars. Analysis was various ensemble decision tree models using SAS Enterprise Miner.

Heterogeneous-Ensemble-Models

https://github.com/melissaEYE/Heterogeneous-Ensemble-Models

The objective of the analysis was identify bad car buys for a second hand car sales business in order to help them avoid cars that were bad and develop an inventory of quality cars. Analysis included various ensemble Support Vector Machines (SVM), different forms of decision trees, regression, neural networks, and Naïve Bayesian models using SAS Enterprise Miner.

Support-Vector-Machines

https://github.com/melissaEYE/Support-Vector-Machines

The objective of the analysis was identify bad car buys for a second hand car sales business in order to help them avoid cars that were bad and develop an inventory of quality cars. Analysis was conducted using various SVM parameters in SAS Enterprise Miner.

Google Colab & Python

Deep-Learning-Convolutional-Neural-Networks

https://github.com/melissaEYE/Deep-Learning-Convolutional-Neural-Networks

The purpose of the analysis was to accurately identify objects in the images, such as a truck or car. To accomplish this purpose, a method of deep learning referred to as Convolutional Neural Networks (CNN), was employed.

Reinforcement-Learning

Reinforcement Learning (RL) is one of three machine learning paradigms (supervised and unsupervised learning are the other two areas). In RL we are concerned with how "agents" or robots take actions within the environment.

https://github.com/melissaEYE/Reinforcement-Learning

Tic-Tac-Toe

The purpose of the exercise is to create a model that learns how to play the game Tic-Tac-Toe. There are two versions of the game, the first, is a 3-x-3 matrix and the second is a 4-x-4 matrix. Original code is credited to Ryan Rudes (2020).

https://github.com/melissaEYE/Reinforcement-Learning

Robot Factory Navigation

The purpose of this assignment is to train a robot to navigate an environment (a guitar building factory) such that the robot moves through the factory in the most efficient manner to collect the required parts for building the guitar. To achieve this goal requires that the robot learn the factory layout (environment), the rooms or locations within the environment (states, n = 9) and how to navigate them (actions).

https://github.com/melissaEYE/Reinforcement-Learning

Big Data

Big database architecture w/ Hadoop & Hbase

The purpose of this assignment was to create a Hadoop HBase database with academic records.

https://github.com/melissaEYE/Big-Data-Hadoop-Apache-Spark/blob/main/Hunfalvay_Assignment_1.docx

Regression w/ Apache Spark

The purpose of these assignments was to use the big data platform to reun various analytical processes.

https://github.com/melissaEYE/Big-Data-Hadoop-Apache-Spark

Sentiment Analysis w/ IBM Db2 on Cloud

The purpose of this assignment was to understand sentiments on airline tweets using IBM Db2 on Cloud service on IBM Cloud.

https://github.com/melissaEYE/Sentiment-Analysis

Text Analysis w/ IBM Watson Studio

The purpose of this assignment was to analyze tweets from a GOP debate using in order to provide insight into Mr Trump's win. I used DB2 on Cloud service and the integrated RStudio environment in IBM Watson Studio.

https://github.com/melissaEYE/Text-Analysis

Chatbot using IBM Watson

The purpose of this was to develop a chatbot for the Convergence Insufficiency Symptom Survey to improve business efficiencies. You can test the chatbot here: https://web-chat.global.assistant.watson.appdomain.cloud/preview.html?region=au-syd&integrationID=eeea12e5-f8bf-4bd7-9a97-748b1586b6eb&serviceInstanceID=5a8de2b0-318a-4570-a621-ac68c4804ef0 If you are interested in further development of this for your business please email me: melissa@righteye.com

https://github.com/melissaEYE/Chatbot-Development

Capstone Project: Eye Movements Across the Lifespan

A Capstone Project is a multifaceted body of work that serves as a culminating academic and intellectual experience for students. My Capstone project included 262,000 rows of data and 1,319 columns. Data included eye movement tests using the major eye movements. Coding languages included SQL, Python, R and json. Auto ML tools include Data Robot. Analytics included: K-means clustering, Affinity Propgation, mean shit, spectral clustering, Ward hierachical clustering, agglomerative clustering, DBSCAN, OPTICS, Gaussian Mixture methods, BIRCH and Bisecting K-Means. Linear regression was used to inform input variables. Statistical tests included MANOVA, Wilks' Lambda, post Hoc LSD & Tukey HSD. Where appropriate some models like the Affinitiy Propagation required Chi Sqared and Likelihood ratios. A successful model was developed identifying 15 age groups from 6 to 80-years-of-age.

Presentation 1: https://www.youtube.com/watch?v=IblaK_UAjbI

Presentation 3: https://youtu.be/mvRwi3c6Tfo

This work is proprietary to RightEye, LLC. To request further information please email me: melissa@righteye.com