1v1 Predict accurately predicts the winner of a League of Legends 1v1.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
css
img
js
LICENSE
README.md
error.php
index.php
matchup.php
riot.txt

README.md

1v1 Predict

####LIVE DEMO AT http://pebocomputers.com/Riot/

Welcome to my entry for the April 2016 Riot API challenge.

This project aims to make an accurate and reliable prediction of who would win a 1v1.

This prediction uses data on:

  • K/D/A
  • CS
  • Winrate
  • Rank

...but specifically uses the data for your most mastered champion.

Your rank will also be included in the prediction. Ranked tier is compared, and if the tiers are the same (Silver = Silver) then the divisions are compared (Silver II =/= Silver V).

The 1v1 Predictor creates a weighted comparison of two summoners.

This weighted comparison allows us to determine the winner of a 1v1.

The comparison is weighted so advantages in categories are worth different amounts.

In order of greatest weight to least weight:

  1. Ranked Tier

  2. K/D/A ratio

  3. Creep Score

  4. Winrate

  5. Ranked Division

More specifically, it's weighted so that you can get a certain number of points in each category that is compared. Each category has a max number of points that can be attained, and is weighted that way, too.

The max number of points attainable for each category is below:

Category Points
Ranked Tier 310
K/D/A 127
Creep Score 102
Winrate 76
Division (if applicable) 35

Now, let's talk about some of the technical stuff.

How do we fetch data about summoners?

Using PHP, we query the Riot API, using the first summoner provided. We check for empty information occasionally. (In matchup.php)

  • First off, we use session_start(); to store error data. More on this later.

  • Next, All functions to retrieve data from Riot are defined.

  • We retrieve Summoner ID, checking for emptiness.

  • Next, we check what champion they have mastered. 1v1 Predict grabs their most mastered champion.

  • After that, we check their ranked winrate with that champion, and convert it to a percentage.

  • Next, we grab K/D/A for that champion, and round it to the nearest hundredth.

  • Creep Score is calculated next, and rounded to the nearest whole number.

  • After that, Ranked Tier and Division are grabbed.

Error data with session_start();? Emptiness checks? How do those work?

PHP's session allows for storage of data across pages. If a crucial element of the comparison is detected to be empty with empty(), session stores the error message in the browser, temporarily (matchup.php). The user is then redirected to an error page, where the error is read from the browser storage and then written on the page (error.php).

How are you comparing Ranked Divisions and Tiers? Those aren't numbers!

Ranked divisions and tiers are NOT numbers. In order to compare them easily, an enum should be created. Unfortunately, PHP has no native enum functionality, so it was replicated through abstract class constants (matchup.php). We assign divisions and tiers numbers so we can compare them easily. Tiers make sense, but divisions are counter-intuitive, since a Roman Numeral 5 is actually a zero here, while a Roman Numeral 1 is a 5.

  • Bronze = 0

  • Silver = 1

  • Gold = 2

  • ...etc...

  • V = 0

  • IV = 1

  • III = 2

  • ...etc...

Then we can compare: constant('Tiers::Bronze') < constant('Tiers::Gold') and it will return True.

Likewise, we can compare: constant('Divisions::V') < constant('Divisions::III') and it will return True.

Setup/Usage

Setting this up on your own is very easy. After installing PHP, clone this repository. Place the repo files in a place that PHP can serve them. You'll also need to replace <API_KEY_HERE> with your actual API key from https://developer.riotgames.com in matchup.php. Everything else should work out of the box. All dependencies are included.