Prediction with categorical and continuous Variables

I want to predict the result of a match in a video game (win or loose).

It's 5 players against 5 players game, who each plays a specific character.

• I have : the ID of each character (there are 150 different characters, so it's a number between 1 and 150)
• I have different statistiques about each character (attaque, defense : number between 1 and 10)
• I have the different ID of the two abilities chosen by each players (player 1 chose abilities number 1 and 3, player 2 chose abilities number 2 and 3, etc ...)
• I have the number of game played by each player, and the number of game won by each player.

At the end, I have categorical variables (character and ability IDs), but also continuous variables (number of game played by players, etc ..)

I have two questions :

1 - How should I do to handle these two types of variable ? I tried KNN and Decision Tree. But KNN doesn't handle categorical variable (so I tried with a vector of dummy variables, but how should I scale my data ?) And decision tree doesn't handle continuous variables.

2 - How can I integrate the fact that in my input vector, a specific variable (like one ability) is correlate with a specific character variable. Don't know if is clear, but imagine I have 30 columns as an input (10 = 5+5 characters, and 20 = 5*2+5*2 abilities). How can I integrate the fact that the abilities in columns 11 and 12 is linked with the character in column 1 ?

If my question is not adapted for with website, I will be very grateful if you could tell me where I could get an answer !

Thank you