Description: This Research Assistant app could generate a short report about any topic.

Description: This notebook shows how to perform sentiment analysis and perform summarization on long document.

Description: This notebook shows how to use pre-trained model and fine tune it to get higher accuracy. It serves as a reference for my further deep learning projects.

Description: In this project, I perform a survival analysis using Germany life table from year 2008.

Description: In this Medium article , I describe how neural network works. I also show how you could create one from scratch using Python Numpy library. Furthermore I also show how to create one using PyTorch.

Description: In this Medium article , I describe only the “Data Collection” step for my text analysis project. In this project, I used LDA machine learning algorithm to discover recurring topic of 260.000 text documents and its trend.

Description: This article is about how to use price elasticity theory to determine optimal price that maximizes revenues.

Description: In this project, I collected ~260.000 text data to figure out the current trend of security-related topics. Due to huge amount of text data, an unsupervised topic discovery algorithm is applied, i.e. Latent Dirichlet Algorithm (LDA). Before applying LDA, it is necessary to preprocess text data and convert it to be compatible as input for LDA. Furthermore I also did data exploratory analysis to figure out the general trend of security topics and general other programming topics.

Description: I had a programming project that needs to read a bulk amount of insider transactions (form 4s) from sec.gov daily and build a rank list of the most successfull insiders in the US. Therefore I need to build a Python script that could make millions URL-requests efficiently, remove unneccessary form 4s and evaluate the remaining datas as Pandas DataFrame. In this Medium article , I will only concentrate on the downloader programming part.

Description: This project aims to help calculating the fair value of a company stock. It calculate if a stock is over- or undervalued, based on its last available annual earning and current interest rate. Without this web app, the author would have to scour the internet to copy the financial datas and entering them manualy into calculator to get an estimation. Warning: If the app is in sleep mode, it would take 30–45 seconds to reactivate, so have patience.