Multi-Task Learning, a Sentiment Analysis Example

The article explains the concept of multi-task learning using an example of aspect based sentiment analysis. This POC solution can be used to generate rating scores for pre-defined aspects of real world product or services out of the customer reviews.


Image by Harold Taylor

From Python to Julia: Feature Engineering and ML

In addition to data manipulation and EDA, Julia is also equipped with a full set of tools for conducting predictive modeling, including both traditional ML and deep learning. In this article, I'd like to carry on my exploration of Julia mainly to test out its support for the modeling tasks.


Photo by CardMapr.nl on Unsplash

From Python to Julia: Basic Data Manipulation and EDA

As the next generation of the statistical programming language, Julia has been gaining spotlight in recent years. Is it ready to replace Python to become the go-to language for the data science practitioners? In this article, we are going to leverage the resources offered by Julia, to tackle a few basic data manipulation and EDA tasks most data scientists would experience in their daily work. I'd like to share my "user exeprience" as a newbie of Julia.


My Observations of the Current Data Science Industry and Career Advice - an SEA point of view

How much "science" work is there within a data scientist's job scope? The title of "data scientist" is getting inflated, and the value of the DS work varies across companies serving different services and product. As a junior DS, I'd like to share my understanding of the current DS and analytics industry and offer some advice to the ones who are keen to pursue a DS career.


Crack Site Selection Puzzle by Geospatial Analysis (Part 2)

In the second blog of the site selection series, we are going to investigate the existing supply of the grocery outlets in Penang, and uncover the imbalance between the demand and supply. It is recommended to build outlets at locations with large gap of demand and supply, in consideration of its potential to allow the client to generate higher revenue.


Photo by author

Crack Site Selection Puzzle by Geospatial Analysis (Part 1)

An end-to-end geospatial application built on open-source tools and data

Site selection is a popular topic in the consulting industry. In this blog, I am introducing a novel solution backed by open source OpenStreetMap dataset and QGIS. Hope this case study can bring some fresh ideas to the analysts and consultants who work on the site selection related projects.