Andrew Scheiner’s Website

I Made My Own Top 100 Mlb Prospects List

2022-05-21T00:00:00+00:00

5/21/22

Hi readers! Long time no see (again I guess)!

Today I want to share an MLB project I have been working on for a few weeks now. For those of you familiar with the MLB, you might know that MLB.com has their very own "Top 100 Prospects List" listing the best up and coming baseball players in the league. Here is a link to their list: https://www.mlb.com/prospects/top100/. Over the past year or so, I have become a big fan of paying attention to prospects and trying to see who the next big stars are. However, I wanted to come up with my own Top 100 list using Minor League Baseball player data provided by Fangraphs (Fangraphs' MILB "Big Board") and my programming experience in R. This post will describe my process. Enjoy!

[To be continued]

-Andrew

I Wrote My First Data Analytics Report

2022-05-21T00:00:00+00:00

5/21/22

Hi everyone! It's a busy day for Connecting the Game as I have been working on a lot of different projects over the past few months and really want to share them.

I recently finished my first class in data analytics, an introduction to data science course going over the R programming language and several statistics concepts related to data analysis. For our final project, we had to write a report on a dataset that we were interested in and use the modeling and analytical techniques we learned to process, visualize, and interpret our data. I chose to analyze the 30 MLB franchises from 1998-2021 and try to find similarities and groups between the franchises while also using their performance to possibly predict how well an expansion team might perform in today's MLB.

I would write a whole big explanation about my report, but I think I should let it speak for itself. To check out the final report, use this link: Data Analytics Papers. I also used supplementary GitHub repo for this project which includes some explanations, outputs, and the dataset I used for this project. Here is that link: Analyzing MLB Franchises Since 1998

I really thought that the outline I used help organize the paper and make a solid foundation for how to approach a data analytics problem. This was helpful especially because we had to use both unsupervised and supervised learning (I go into more detail about them in my report) and discuss modeling and results.

I hope you enjoy!

-Andrew

Responding To A Claim About Data

2022-05-21T00:00:00+00:00

5/21/22

Hey! Another post, look at that. This is the last post I will be starting for the day, but it might be my most interesting one.

Much like how I just finished my introduction to data science course (see my previous post regarding my data analytics report), I also just finished a very interesting course in the philosophy of data. It was definitely the most interesting class I have ever taken, and I learned a lot about the philosophy of science, making an argument, and the philosophy of data. Now, this topic is essentially data ethics, however, the class focused heavily on the history of science and how to generate an argument, objection, and replies to objections using passages we read. For our final assignment, we were tasked with taking a passage from Kate Crawford's popular article, "The Hidden Biases in Big Data," and responding to it.

I would love to write a whole post really diving into this topic, but I already did it in my paper, so I will let it speak for itself.

To check it out, here is the link: Data Analytics Papers. I have included Kate Crawford's article on there too!

Enjoy!

-Andrew

Predicting Mlb Teams’ Success

2022-01-12T00:00:00+00:00

1/12/22 (*Work in progress*)

Hello readers! Long time no see.

After taking a long break dealing with more pressing matters, I finally found the time to work on the things I now really love doing, working on computer science projects such as this one. While I have been working on improving this site of mine, I also created a brand-new web site, an interactive web application which deals with machine learning and Major League Baseball. The name of the web site is Predicting MLB Teams' Success. This article will dive into my reasonings for making this project, the process I underwent to create it, and my final thoughts.

Why did I make this project?

As I am now on a nice break and wanted something to re-kickstart my work in the field of computer science, I chose to set out on a quest to find a better format for creating websites using my favorite software, Visual Studio. With the 2022 edition out and the creation of .NET 6.0 (basically 5.0 with long-term support), I decided that learning a new method for web development would be the best bang-for-my-buck to relearn. In my quest, I discovered Blazor, a Microsoft-based web framework that uses C# and HTML coding to create pages and components for a rich user interface (Microsoft). So, now that I found a development platform, I needed an idea. As you may have seen from some of my earlier posts, I have a fascination with the MLB and baseball as a whole. When I was introduced to machine learning and Visual Studio's Model Builder using ML.NET, I started looking for ways to take baseball data and generate model to make accurate predictions. My first true attempt to use machine learning with baseball data was to analyze 2021 MLB umpire's performance to determine if the home or away team would be favored in a given game. Even though I had created a structured and interactive C# console app to accommodate my data, I realized in the end that the model was only guessing its predictions. Since the r-squared value on the dataset was around 0, there was no correlation in the data and so any prediction the model made could not have been accurate.

Well, time passed, and my next idea sparked. I would attempt to predict how many games a team wins in a season based on a team's total Wins Above Replacement (WAR) statistic. WAR measures how much better or worse a player is than their replacement, or in other words, a player's total contributions to their team (The Stadium Reviews & Fangraphs). Yes, WAR, might not be the most accurate stat for measuring how good a player is, yet combining every player on a team's WAR stat does have adequate correlation with success. For any MLB team, success can be measured in several ways, whether this is making the playoffs, winning the World Series, or just getting positive results even when rebuilding. However, for me, success for a team comes from how many wins they will achieve. I hope this project will allow users to learn how well their favorite team might do based on how good their players are.

The process of making this project

When I first started working with machine learning and baseball data, my goal was always for there to be an interactive application for users to gain insight through prediction. My umpire performance model was unavoidably unusable because of the dataset, so I turned my attention over to predicting wins. My initial idea for this projects' purpose was to "Use historical MLB team win and WAR data to predict any MLB team's win total based on projected/actual total team WAR." So, the process of making this project began desiring a simple console app which would simulate individual teams' seasons for projections or simulate entire MLB seasons based on projected WAR. Additionally, three main variables were potent in using my data and model together: knownWins, expectedWins, and knownWAR. And of course, to understand the data I used, I went to Baseball-Reference and manually entered each team's total WAR data from the years 2006-2021, excluding 2020. I chose this particular set of data because MLB experts have labelled this time period as the "post-stereoid" era. I assume most users will use this website to make future predictions for their team, rather than try to rethink the past, so I wanted results to be as realistic to today's era of baseball as possible. In addition, I excluded the 2020 MLB season because each team only played 60 games, which could create confounding results. I already had prior experience working with ML.NET and C# Console Apps, so this intro phase did not take very long for me, and I then quickly turned to the more difficult part of the process: converting my idea into a website.

Instead of using ASP.NET Core 2.1 like I did for building this blog (it is now out of support), I took the time to learn Blazor Server, the framework I mentioned above.

[To be continued]

-Andrew

Becoming An Automated Baseball Statistician

2021-05-28T00:00:00+00:00

5/28/21

Hello readers!

Sorry it has been so long since I posted last. May was very busy.

This is honestly the first post that truly encapsulates what I will be using this blog for: merging sports and computer science. Today we will look into the hot topic of baseball stats and not only how they are calculated, but more importantly, how they can be calculated with efficiency. Let's just say it is not enjoyable to manually calculate some baseball stats, but we can do so automatically using Python programming and existing data available on the Internet!

Here we go! Let's outline the steps necessary to become an automated baseball statistician.\

Set up a workspace to use our data as input into a program
Collect existing baseball data (we will be using data from the 2019 MLB Season found on https://www.baseball-reference.com/
Find formulas to calculate our desired stats\
Create a program that reads our data (input) and calculates then outputs our desired stats

Setup a workspace

Whenever I work with programming like we will doing today, I always work in Visual Studio Code (VSC). Much like its friend Visual Studio developed by the same company, Microsoft, VSC specializes in coding and even has really good development and extension features. I recommend downloading it to make a workspace where our data is easily accessible: https://code.visualstudio.com/Download.

Once you have downloaded VSC, if you do not have Python installed already, please use this link to download its software: https://www.python.org/downloads. I recommend downloading the latest version available. If you are unsure, just download a 3.7 or 3.8 version.

Create a folder on your computer for this project and if you right-click on it, you should get the option to "Open with Code". Click on this to see your folder open in VSC's folder explorer on the left-hand side. You will want to make 2 new files right away in this folder.

batter_data.csv
pitcher_data.csv

Click the "New File" button next to the folder name to create a new file. This folder will eventually have 6 files total: two more .csv and two .py files from GitHub.\

::: {.tss_actions} :::

Collect existing baseball data

For this article and teaching purposes, I will be using two sets of customized data from Baseball Reference. These are the two sets:

2019 MLB Standard Batting Stats: https://www.baseball-reference.com/leagues/MLB/2019-standard-batting.shtml
2019 MLB Standard Pitching Stats: https://www.baseball-reference.com/leagues/MLB/2019-standard-pitching.shtml

I used these two data sets and customized them for the purpose of this article. I did so by clicking the "Share & Export" dropdown menu and choosing "Modify, Export & Share Table". This allowed me to exclude^1^ certain columns so that we could manually calculate the stats. To make this data usable for our Python program, we must use the same dropdown to choose "Get table as CSV". Once the table converts to a text format on the website, copy and paste all 1000+ rows into the .csv we created in the section above based on which table you collected from. The batter data from Baseball Reference goes into the batter_data.csv file and the same with the pitcher data.

Good job! You just learned how to do simple data collection. Note that if want to present this data in any form to reference the source you got it from. This applies to anything you find on the Internet!

For this article, I changed some things up so that your program only analyzes around 10 players at a time instead of all 1,000. You can download those "partial data" .csv files I used through my GitHub: https://github.com/Baseballfan5303/VS-BlogWebsite/tree/main/BlogArticles/AutomatedBaseballStats. Using only a part of the large dataset allows testing the Python program to be quicker and easier.

Finding formulas

Before we go through the code, we need to find the formulas to calculate our desired stats. The website we will use for this is Fangraphs (https://library.fangraphs.com/). Use the search bar to type in each stat in order to find the proper formula to use.

Creating the program

I have added the code to both the batter and pitcher stat programs in the same GitHub location as above. Please feel free to download the .py files and use/test with them; you will want to save the Python files in the same folder you created in step 1 and added the .csv files to. You might be wondering how I came up with this program what each section of the program means, and what we are getting as output. Well, let me explain.

Batting Stats:

We want to find batting average (AVG), on-base percentage (OBP), slugging percentage (SLG), on-base + slugging (OPS), batting average on balls in play (BABIP), total bases (TB), and isolated power (ISO).
First, we import csv so we can read a .csv file.
We then define our variables which are the stats we want to find. singles is a global variable because we need to calculate this stat for TB.
The playerStats array contains each batter's stats as we iterate through our .csv rows. Each index value of the array represents a stat column.
The function open(file, mode) in Line 22 allows use to read our partial data .csv file. Make sure the file we are opening is the "batter_partial_data.csv"!
Line 23 sets myData as the .csv reader; myData represents the rows of the sheet.
Line 24 starts our for loop iteration through the rows of the sheet. The playerStats array becomes rw in Line 26.
Lines 29-37 set the variables our formulas require by referencing an index value of playerStats based on the column headers in the csv. For example, ab = int(playerStats[7]) means at bats = the integer value of the 8th column in this row of data.
In lines 40-50, we perform our calculations. Setting our variables in Lines 29-37 allow us to change index values for our formulas in only one place, instead of multiple.
For batters, decimal stats are rounded to three decimal places, hence the round(...,3) function being used.
Line 53 prints the batter's name in the row we are currently iterating through.
Lines 54-60 print the batter's stats.

Pitching Stats (anything listed as explaining the pitching stats program differs from the batting stats program):

We want to find earned run average (ERA), adjusted earned run average (ERA+), fielding independent pitching (FIP), walks + hits per inning (WHIP).
When defining our variables, we need the league ERA in 2019, lg_era (4.49 in 2019), for ERA+ and a constant for the FIP formula, fip_const (generally 3.1).
Be sure to use the "pitcher_partial_data.csv" file in line 20!
Notice that we are getting different variables and index values in lines 27-33. This is because the formulas for our desired pitching stats are different from the batting stats.

Voila! You have now become an automated baseball statistician! We used data in the form of a .csv, ran it through a Python program, and got desired statistics outputted!

I encourage you to play around with different sets of data, try to calculate even more stats, and even try to repeat this process for another sport! Remember to manually change the file being read in each program and any constants that might change from year to year.

Good luck with your data collecting and programming!

-Andrew

^1^ Excluded from batter data I collected: AVG, OBP, SLG, TB; Excluded from pitcher data I collected: ERA, ERA+, FIP, WHIP

My Live First Thoughts On Mlb The Show 21

2021-04-23T00:00:00+00:00

4/23/21

Happy Friday everyone!

I thought I would introduce the sports aspect of Connecting the Game with a classic video game franchise that is regarded as one of the best in sports video games. No, not Madden...MLB The Show (MTS). MLB The Show 21 released for everyone on April 20th for both Xbox and PlayStation for the first time in its existence. My history with baseball is that I have been watching it for twelve years, I play for teams myself, and I am a diehard Texas Rangers fan. My history with baseball video games starts all the way back at MLB 2k10 on the Wii and MLB 2k13 on the Xbox 360, with my most recent game being MLB The Show 19 when I picked up a PlayStation. So, now that you know where I am coming from, here are my live first thoughts during my first time playing the official new MLB The Show 21. (I used a stopwatch to get the near exact minutes and seconds for fun!)

(1:00) Great intro!

(2:00) I like that they give you the option to play one of three different play styles but let's see how they are. I will start with Simulation.

(2:30) Time for practice!

(2:50) I have to face the Dodgers as the Rangers in the practice game

(6:20) Oh boy is pinpoint pitching gonna be interesting.

(7:30) Ok I will try track fielding for now...

(9:30) Wow I already got an achievement? Just for breaking the hitter's bat.

(9:50) First perfect pitch of the year!

(11:50) So I finished my first half inning in the demo game. Thoughts so far:

Pinpoint is going to take a while to get used to; much different than Xbox 2k13
Nope, track fielding does not work for me. I need to have that ball icon.
The graphics seem laggy on the Xbox One maybe because of framerate limits?

(13:50) I love Globe Life Field.

(14:40) Lol Khris Davis is hitting 3rd in this lineup.

(15:30) I like the background beats.

(16:45) Ok, I'm done with the demo.

(18:00) Ooh, a free player for playing the Tech Test. (It's Gio Urshela...)

(19:10) I am giving the main menu design and layout a 10/10.

Slick and organized. I'm glad they do not try anything too special or crazy.

(20:00) None of the soundtrack songs sound familiar to me.

(20:30) So we have 4 featured game modes: Road to the Show (RTTS), Diamond Dynasty (DD), March to October, and Franchise.

Then there are 4 other categories below: Quick Play, Play Now, Other, and Learn

(22:00) Time to build my ball player!

(55:00) Gosh that took a while to make. I tried to make myself and did not like it so I started over and randomized. Onto loadouts!

They make no sense.
Why can’t I just set my primary and secondary positions?

(59:00) Onto Diamond Dynasty!

(59:50) Income the Texas Buffaloes.

I don't know why that was the team name that came up, but I love it.

(62:20) Nomar Mazara with the walk-off single. First Moment of MTS 21 done.

(65:40) Brad Wieck gets my first save.

(70:40) I feel like there is a great balance between CPU and online content in Diamond Dynasty.

(74:10) Into RTTS I go. I would like to play for the Rangers, but wherever I go is cool.

(75:20) Cool RTTS intro, but where is the pre-draft stuff?

(By stuff I mean the scouting games and drills pre-draft time)

(76:00) I get drafted by the Rangers.

(76:20) Lol the MTS Player's League guy is back for a generic draft pick intro.

(82:10) First game in the minors: Played DH and went 1-3 with a double. Franchise mode time.

(91:00) So, no Kohei Arihara? There is a pitcher named Rick Slade on the Rangers but he doesnâ€™t exist.

And who is Rodney Pressley supposed to be?
Where is Anderson Tejeda?
(I started with a Rangers franchise, of course)

(94:00) No changes to the Franchise mode starting options. But hey, at least I can make a league-wide DH rule and no extra-innings runner like there should be.

I eventually took a break and have played about a dozen hours since launch day, so here are my now final thoughts from my first few experiences:

I don't know if I am the only one feeling this way, but I felt that the framerate of MTS on the Xbox One was not good in the practice game nor franchise. It felt like I was constantly behind in animations, making hitting very difficult.

Yet, when I when into a mode like RTTS, it seemed that the gameplay was smooth and that there were never any framerate issues at all.

This honestly might just be because of the crowd.

Upon continuing the start of my Rangers franchise, I decided to go straight into rebuild and trade away as many C and D potential players as I could for prospects like Wander Franco, Adley Rutschman, and MacKenzie Gore.

The trading system is either too legit or way too broken.

On second thought, trading is broken, but I love it and recommend to everyone starting a new franchise to just go for some cool 65 overall A potential prospects.

After playing a lot of MTS 19 on PlayStation but skipping MTS 20, I really like the hitting feedback changes, but I find the hitting to be more difficult due to the PCI when zone hitting. As for RTTS, I like how you can see live how your player is progressing as they have plate, pitching, and fielding appearances.

I got the chance to play the Tech Test in February which has been very helpful as I was kind of already adjusted to the game, but I knew what to look for. Excited to get to play with some of the legends I missed out on in MTS 20.

The changes the made this year have been minimal but have felt big, the one that sticks out to me manager-wise is the new addition of a depth chart in franchise to forecast the future of your team. Gameplay wise, the big change is the revival of pinpoint pitching. I played hours of MLB 2k13 on the Xbox 360, the most recent MLB game made for Xbox (sorry RBI Baseball, you don't count), and pinpoint pitching is the main option for pitching controls, so I am using it and it has been pretty fun with some interesting results.

Last but not least the features and feedback of things I do not like and/or need improvement. Here they are:

A desperate need for framerate increase on the Xbox One. I don't have the Next Gen consoles and do not intend on buying one anytime soon, so I am stuck with the worse version of the game. This is fine except for the fact that my gameplay almost always feels laggy with low frames. I don't know if this is because of my console being 5 years old or my TV, but I assume not. Please let me know if you have any idea why this is happening. (preferably is you play the game on the older consoles)
Stadium creator is a Next-Gen exclusive...stupid.
Missing early by about 0.1 seconds on a pinpoint pitch timing will make the pitch an automatic meatball, so I feel something needs to be done so that when I aim low and away, a slight bit of mistiming does not result in a grand slam.
As for loadouts and the new Ballplayer structure, it all seems too confusing upon first glance. In MTS 19, I was able to select my primary and secondary positions explicitly, but now I can't? I get the idea of being a two-way player but I can only play one position as of right now instead of the many I wish to be eligible for.

Hopefully you had a good laugh while reading this and if you love baseball and video games, I absolutely recommend playing this game.

Until next time!

-Andrew

Beginner’s Guide To Computer Science

2021-04-17T00:00:00+00:00

4/17/21 (*Work in progress*)

Welcome back everyone,

This post will be outlining how to begin learning about computer science and what you can do with it. While this post says beginners, people of all levels of computer science knowledge are encouraged to keep reading this article to get an idea of resources and strategies that are available for free online.

So, this guide will outline the following topics (italicized = done):

What is computer science?
How big computer science is and application vs. theory
Where to get started if interested in computer science
How to find online material for guidance
Where to find help when stuck
How to prepare to be a computer science major in school

What is computer science?

Well, great question. Merriam-Webster defines computer science as "a branch of science that deals with the theory of computation or the design of computers." And Wikipedia gives a much more detailed definition: "the study of algorithmic processes, computational machines and computation itself pans a range of topics from theoretical studies of algorithms, computation and information to the practical issues of implementing computational systems in hardware and software.” As someone who has been developing and programming for fun over the past few years, I feel like the simplest definition I can give is that computer science is the study of how computers function and the endless possibilities that come along with them. By saying "endless possibilities" I am hinting at the potential computers have and how many different aspects of computers are used in many different specialized fields.

How big is computer science? (& application vs. theory)

Computer science covers a very large span of specialized fields in which you can start learning about today. You have two different general areas in which all of the specialized fields generally fall under: applied and theoretical computer science. Theory for computer science differs from theory in other studies because of how important to learn new methods and strategies to be implemented in development of projects. Topics include: Algorithms, Data Structure, Computation, and more. When we talk about applications of computer science, however, topics of study and fields are much more specialized than the theory necessary for application. Topics of application include: App/Web Development, Cybersecurity, Networking, Programming, Data Science, Machine Learning, and lots more. From my experience, this is how wide I have reached with computer science after learning in school and working on projects/jobs, but let me show you what Britannica has to say about how big computer science has grown.

Britannica talks about computer science having traditional studies like Computer Architecture, Programming Languages, and Software Development. But what I like that they mention is how computer science intervenes in other fields of study such as biology to form Bioinformatics and chemistry for Computational Chemistry. So we can use computer science knowledge and fundamentals to help learn and innovate other major fields of studies like physical and biological science. Here are the 15 distinct fields Britannica mentions make up computer science:

Algorithms and complexity
Architecture and organization
Computational science
Graphics and visual computing
Human-computer interaction
Information management
Intelligent systems
Networking and communication
Operating systems
Parallel and distributed computing
Platform-based development
Programming languages
Security and information assurance
Software engineering
Social and professional issues

Where to get started

Let's say you are like me, someone who has some background in computer science from school, but not much knowledge on where to get started learning what you want on your own. This is what I recommend you do:

Decide which field(s) of computer science interest you.
Don't just Google Search "how to get started with" because it will give you too many results. Head to Microsoft Learn before you resort to using your search engine: https://docs.microsoft.com/learn.
Once on Microsoft Learn, create a Microsoft account if you don't have one or sign in if you do have one. (Not entirely necessary, but good if you want to save what you do in the next few steps)
Click on "Customize my path" so that the "What roles are you interested screen" shows up with a bunch of different options to click.
Go through this quick survey so that Microsoft will be able to give you the best learning modules to start working on. I recommend choosing Visual Studio as your product if you are unsure.
Click on "Browse recommendations" to see which learning modules you were recommended to start with.
When browsing your results, pay close attention to the color, time, and tags of each module. And, of course, be sure to read the description/title. Color signifies product and knowing what software and time will be needed for the module is important in this process, so be aware as you go through your results.
Pick any module for now to look at.

Upon clicking on the module, you will see more tags and descriptions of what you will be learning, as well as learning objectives. But the most important part is the Prerequisites you need to follow along. For a Xamarin module, I need to have Visual Studio 2019 installed and be familiar with C# and .NET. So, before you get started with any of these modules make sure you know what you will be learning, how long it will take you, and what software/prereqs you need to continue. All of this will help you get a great head start on your journey to learning something new in computer science.

[To be continued]

-Andrew