As I continue with job searching during these times, it appears that I should expand my knowledge of other programming languages. Some entry-level positions require such knowledge along with SQL. Seeing that SQL is something I have learned already, perhaps I can record my journey of learning Java here. Previously, I had already made a post about Java, but I was rather Clueless as to what most things meant. However, this time I believe I have a better grasp.

Basics

  • Class: A template used to instantiate Objects. It determines what states and behaviors the…


It’s been a while since my last post, as I’ve been working on various things. However, some progress has been made on my project since then. I have added pmdarima and LSTM network to the time series forecast, and I think they provided rather interesting results.


This dataset is not as easy-going as the player in this image. Image from Dev Blog

As I continue through this project, the main thing that sticks out to me is that this time series is very unusual due to the spike in number of reviews in the most recent year. As I will show later in this blog entry, this makes it very difficult for forecast modeling to accurately predict the number of reviews as time goes on.

Because the time series looks so unusual compared to most datasets I’ve seen, yet is probably typical of how some games are, I decided that my main priority for this project is to figure out the best…


Daily reviews for each year is graphed. Bottom rightmost graph is all the data.

Continuing off from before, one thing I noticed is that because the plot of the data encompasses so many data points (daily) over the span of about 7 years, it makes sense to separate the data into each year in order to get a better understanding.

Immediately I see that during the beginning of the game, there are rather rocky peaks in the data probably due to frequent patching and subsequent feedback. After a somewhat active 2 years, the game enters a rather long period of time where there are very few reviews every day. However, for the next 5…


This is actually a follow-up to the CS:GO review analysis. The main issue before was that scrapy spiders constantly stopped for no obvious reasons before obtaining a significant amount of reviews. In the end, I decided to use Steam API. Luckily, someone has written code to get steam reviews for any game using the Steam API.

Initially, I tried to use Steam API to get every single review for CS:GO. …


It’s time to begin another project of interest. During my time at Flatiron, I had analyzed Steam games from a Kaggle dataset to determine what made a successful Steam game. The scope of the project was rather large, and the data was very noisy, thus making it difficult to make any concrete solutions. I decided to focus on something that had substantial data on Steam as well as being something I’m relatively knowledgeable on.

I set my eyes on Counter Strike: Global Offensive, or CS:GO for short. Being an extremely popular game with a long history and large active player…


One of the last things I can think of doing before wrapping up this project for the time being is looking at Apple Store reviews for Ragnarok Mobile Global. For this, I’ll be using this specific scraper in order to get my reviews.

What makes the Apple Store scraper different is that the Google Play one has more information given about the reviews, but the Apple Store one has an input for which countries I want reviews from. Technically speaking, it would be ideal for me to simply input a list of every country possible and iterate the function through…


This will be a relatively short one since it’s the holidays and I had run into some issues with working with a dataset that is 10x bigger than before.

After performing EDA on the SEA region of Ragnarok Mobile, it is clear that there is a lot more meaningful information to be gained from a much larger sample size. This much is obvious. Now to see if the models created from various machine learning algorithms using SEA’s data would be more accurate than that of Global’s data.

Firstly, we vectorize our text with TF-IDF and create a training and test…


One of the obstacles that is holding this current project back is the lack of reviews that can be obtained from Ragnarok Mobile Global. It is also extremely likely that a majority of global players simply do not use the Google Play store to voice their concerns, and as a result, many of the reviews do not have anything particularly outstanding. This may make it difficult to create an accurate model.

There are two main easy options to solve this, which is to either scrape the iTunes store or to utilize the reviews from another server. Luckily, Ragnarok Mobile has…


After spending a lot of time working on other things, it’s time I get back to this project.

After my previous post, I decided to try my luck using a different google play scraper to get reviews, since I only had a few reviews to work with. This one uses Node.js, which I talked about in my previous blog post.

Moses Lin

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store