Pinned Repositories
AB-Testing-Marketing-Advertisement
In this project, the task was to test whether a particular variant of an AD appearing on a client's website performs better or worse than the other variant of the same AD. A/B testing is appropriate here as it will rely on statistical analysis to determine which variant of the AD does better. A/B testing will also give us more objective and statistical evidence that proves which AD is better.
COVID-19-World-DataVisualization
Omicron variant of COVID 19 continues to run rampant throughout the entire world. With just over two years into this pandemic, I'm interested to see some historical data on the speed and magnitude of which COVID has spread throughout the world, and in some certain major countries (e.g. Italy and China). In this project, I use pandas, numpy, plotly and Matplotlib to investigate the past historical data of COVID in the world.
Crop-Disease-Prediction
In this project, I was contracted by Rooftop Republic (https://rooftoprepublic.com/) as part of a Hackathon event to build applications that would bring added business value to their business. With the implementation of technology and improving logistical operations as their focus, I decided to build a AI model that would help them predict and identify the level of heath of their crops, which will help increase their ability to grow crops that are healthy and sustainable.
Insurance-Fraud-Claim-ML-Prediction
In the insurance industry, claims related to fraud is a huge problem. As an insurance company, it can be quite complex and difficult to identify those unwanted claims. In this project, I am working with data provided by an Automobile Insurance company. I am tasked with creating a predictive model, using the data provided, to predict if an insurance claim is considered fraudulent or not. The task is a classification task of Yes/No (i.e. Yes is Fraudulent, No is Not Fraudulent claim). The concluding result is that XGBoost Algorithm with Hyperparameter Optimizing achieved an 82% accuracy with 18% misclassification of fraudulent transactions.
ML-Prediction-on-Employee-Turnover
In this project, it focuses on a real life business problem that every HR personnel and company would face, employee turnover. The data was provided by a multi-national company for strictly to be used for learning purposes, with all personally identifying information removed or edited for the purposes of this project. At the end of the project, I used Random Forests and Decision Trees to design an algorithm to predict the likelihood of an employee leaving the company based on certain characteristics and features.
ML-Telecom-Customer-Churn-Prediction
This project involves predicting the customer churn rate of a telecommunication company. This is important analysis to conduct as Telecom companies would use results from customer attrition analysis as a key business metric. Understanding this metric helps give an estimate of the cost of retaining existing customers, which usually is cost much less than acquiring a new one. In this scenario, Machine Learning (ML) algorithms can help analyze the customer attrition rate. I used several ML models to demonstrate and evaluate which ML model would perform better in this business case.
nchan1994
Config files for my GitHub profile.
thinkcolfiles
Files to work set up
Unsupervised-ML-Bank-Customer-Segmentation
In this project, I was given extensive data on the bank's customers over a 6 month period. In this data provided, it includes the customer's transaction frequency, amount, tenure of being a customer with the bank, and many more. The data is provided in the repository. The data can also be found on kaggle (https://www.kaggle.com/arjunbhasin2013/ccdata)The task is to use the data provided to help launch a targeted marketing ad campaign that is designed for specific customer profiles. Part of this marketing program involves designing ads for 4 specific groups of customers (i.e. Transactors, Revolvers, New Customers, VIP/Prime). This targeted marketing should maximize the marketing campaign conversion rate.
Web-Scraping-on-Dayuse.com
nchan1994's Repositories
nchan1994/AB-Testing-Marketing-Advertisement
In this project, the task was to test whether a particular variant of an AD appearing on a client's website performs better or worse than the other variant of the same AD. A/B testing is appropriate here as it will rely on statistical analysis to determine which variant of the AD does better. A/B testing will also give us more objective and statistical evidence that proves which AD is better.
nchan1994/COVID-19-World-DataVisualization
Omicron variant of COVID 19 continues to run rampant throughout the entire world. With just over two years into this pandemic, I'm interested to see some historical data on the speed and magnitude of which COVID has spread throughout the world, and in some certain major countries (e.g. Italy and China). In this project, I use pandas, numpy, plotly and Matplotlib to investigate the past historical data of COVID in the world.
nchan1994/Crop-Disease-Prediction
In this project, I was contracted by Rooftop Republic (https://rooftoprepublic.com/) as part of a Hackathon event to build applications that would bring added business value to their business. With the implementation of technology and improving logistical operations as their focus, I decided to build a AI model that would help them predict and identify the level of heath of their crops, which will help increase their ability to grow crops that are healthy and sustainable.
nchan1994/Insurance-Fraud-Claim-ML-Prediction
In the insurance industry, claims related to fraud is a huge problem. As an insurance company, it can be quite complex and difficult to identify those unwanted claims. In this project, I am working with data provided by an Automobile Insurance company. I am tasked with creating a predictive model, using the data provided, to predict if an insurance claim is considered fraudulent or not. The task is a classification task of Yes/No (i.e. Yes is Fraudulent, No is Not Fraudulent claim). The concluding result is that XGBoost Algorithm with Hyperparameter Optimizing achieved an 82% accuracy with 18% misclassification of fraudulent transactions.
nchan1994/ML-Prediction-on-Employee-Turnover
In this project, it focuses on a real life business problem that every HR personnel and company would face, employee turnover. The data was provided by a multi-national company for strictly to be used for learning purposes, with all personally identifying information removed or edited for the purposes of this project. At the end of the project, I used Random Forests and Decision Trees to design an algorithm to predict the likelihood of an employee leaving the company based on certain characteristics and features.
nchan1994/ML-Telecom-Customer-Churn-Prediction
This project involves predicting the customer churn rate of a telecommunication company. This is important analysis to conduct as Telecom companies would use results from customer attrition analysis as a key business metric. Understanding this metric helps give an estimate of the cost of retaining existing customers, which usually is cost much less than acquiring a new one. In this scenario, Machine Learning (ML) algorithms can help analyze the customer attrition rate. I used several ML models to demonstrate and evaluate which ML model would perform better in this business case.
nchan1994/nchan1994
Config files for my GitHub profile.
nchan1994/thinkcolfiles
Files to work set up
nchan1994/Unsupervised-ML-Bank-Customer-Segmentation
In this project, I was given extensive data on the bank's customers over a 6 month period. In this data provided, it includes the customer's transaction frequency, amount, tenure of being a customer with the bank, and many more. The data is provided in the repository. The data can also be found on kaggle (https://www.kaggle.com/arjunbhasin2013/ccdata)The task is to use the data provided to help launch a targeted marketing ad campaign that is designed for specific customer profiles. Part of this marketing program involves designing ads for 4 specific groups of customers (i.e. Transactors, Revolvers, New Customers, VIP/Prime). This targeted marketing should maximize the marketing campaign conversion rate.
nchan1994/Web-Scraping-on-Dayuse.com