Let me introduce myself

I am a Data Analyst working in a technology-based company for over a year now, with past experience of working with AI models for industry use cases during my past work.

As a Data Analyst, I usually have to handle to data requests by other departments as I have been the only data analyst in the company since I have joined until recently. Sometimes, I would just provide simple data according to their needs, while recommending some extra metrics that I find useful for their use cases. For more complicated tasks, I would have to plan the metrics to be tracked or used for analysis. Until the end of the tasks, I would provide reports or dashboards along with analysis results & recommendations. To sum up, these tasks would utilize the skills of data analysis, data visualization, machine learning, deep learning, and deployment. I have also written some articles on Medium publications to share some things I have learnt. Scroll down to find out about the portfolio projects I have done on my own. These are not company projects as they are confidential.

Portfolio Projects

Smartwatch Usage Data Analysis

Data Analysis, Data Visualization

A case study done in the Google Data Analytics Certificate program. Smartwatch data was analyzed to provide high-level recommendations to the company. This project makes use of problem solving and data analysis skills to tackle a real-world scenario in a company.
The six phases of data analysis life cycle are incorporated into the case study. The life cycle consists of Ask, Prepare, Process, Analyze, Share, and Act, which begins from ascertaining the business task at hand, to providing analysis results and recommendations to the company.

Photo by Andres Urena on Unsplash

Anime Recommendation Website

Website, Docker, Web Scraping, Deep Learning

A recommendation system based on collaborative filtering and embeddings of user ratings was built using Tensorflow and Keras. Then, a website built with Flask was containerized and served with Nginx server using Docker and Docker Compose. The website was then deployed to a Google Compute Engine's Virtual Machine instance in Google Cloud Platform.
NOTE: The website is only accessible until Sept 2021 due to free trial.

SQL & Tableau Project

SQL, Data Visualization, Tableau

This project actually consists of three different parts. The first part used SQL queries to explore and analyze worldwide COVID-19 data, then the second part used some of the query results to transfer to Tableau for visualizations.
The third part then consists of SQL queries used for data cleaning, including but not limited to handling missing data and removing duplicates.

More Projects

All

Analysis

Visual

Web

Smartwatch Usage Data Analysis

Data Analysis, Data Visualization

A case study done in the Google Data Analytics Certificate program. Smartwatch data was analyzed to provide high-level recommendations to the company.

Photo by Andres Urena on Unsplash

Anime Recommendation Website

Website, Web Scraping, Deep Learning

Creating a recommendation system using Tensorflow. Built a website with Flask and deployed to Google Cloud Platform.
Website accessible until Sept 2021.

Visualization of Malaysia COVID-19 Data

Web Scraping, Data Visualization, Web App

Created interactive visualizations of COVID-19 cases in Malaysia including choropleth map in Python and deployed online with Streamlit.

SQL & Tableau Project

SQL, Data Visualization, Tableau

This project includes exploration and data cleaning using SQL, as well as a Tableau visualization of COVID-19 data.

IMDB Movie Review Sentiment Analysis

Web App, Data Analysis, Deep Learning

Predicting positive or negative sentiments from IMDB movie reviews. PyTorch implementation using BERT model was also tested.
NOTE: The Web App takes awhile to load due to Heroku's limitations.

Photo by Jakob Owens on Unsplash

Used Car Price prediction

Web App, Data Analysis, Machine Learning

Predicting used car prices based on the features of the car. The dataset used consists of many missing and inaccurate data as they were scraped from Craigslist, causing the modelling process to be more difficult.

Photo by Dan Gold on Unsplash