- Predicting potential of a football U23 player, based on popular video game FIFA. Considering age, workout hours per week and current overall rating.
- Predicting car condition over years. Same logic, now considering age of the car and how many times does it get checked by technician per year
I generated players and cars(brands and owners) artificially. However, program still can work with real data
I created corresponding classes with attributes: player with full name, age, workout hours, overall rating and potential; cars with owner, brand, age, checkings per year and condition (let's say every car starts at peak condition --> 100.00 and then decreases) Equations for estimating their potential/condition is made by me. After that I generated training data and created matrix from their properties.
If you run this program you will get a prediction of a custom object and after that see generated predictions for new testing data (about 50 entries)
For the first idea, I use matrix of people to solve (Ax = b) x by normal equation,modified Gram-Schmidt or Householder depending on user's input.
For the second idea, I use matrix of cars to solve (Ax = b) with custom constraint (Cx = d) by modified Gram-Schmidt or Householder depending on user's input.