Testing the AI...Results...Updated! 3/08/09

I've done some testing of the AI in the game for the last month and a half, and I have come up with some numbers based on the formation I was using at the time. The AI tends to attack the weaker unit in the formation than the stronger if it's possible. I tallied up between 3000-5000 attacks for each % listed.

 

Formation used:

Pirate / Lady Knight / Spartan

Northern Viking / Cannon(main) / Witch

Oracle / Monk / Prophet

 

Now here are the percentages of attacks made against each of those columns. I switch the columns around periodically so that if the AI was based on just right, middle, or left column, the numbers would still even out. After this testing, I don't think it matters which column it is, just what mercenaries you have in it. Also I kept track of the magic attacks as well, they didn't seem to deviate too much from normal attacks, so I just bunched them together.

 

Vs. Single Target Melee attack (like sword attack) 5000 Attacks tallied

28.6%/ 39.3% / 32.1% (front line attacked only)

With my main and monk in the same column it was the primary choice of mobs most of the time, but still not that focused on it.

 

Vs. 2 column attack (like spear attack) 5000 Attacks tallied

31.8% / 55.3 / 23.5%

With the ability to hit my main the mobs seemed to go to that column alot more than the other attacks. I did some minor testing and switched my main and monk, and the attack % went back to the "single target melee attack" pattern (only did 1000 attacks of this formation).

 

Vs. 3 column attack (like gunner attack) 3000 Attacks tallied

28.8% / 33.7% / 37.5%

I'm speculating, but due to the witch being a flying mob, the mobs must have picked up that they need to attack the flying mercenaries first.

 

Vs. Single Target Ranged attack (like archer attack) 4000 Attacks tallied

9.9% / 7.4% / 7.2%

8.7% / 12.1% / 13.8%

13.0% / 14.7% / 13.2%

It would seem that with the option to attack any target, mobs would shoot for the weakest mobs first.

 

Vs. Cross attack (like Cannon) 5000 Attacks tallied

I noticed early on that the majority of attacks made by this attack pattern were almost always centered in the back two corners or the center of the formation. I did some extra testing on this attack pattern and found that no matter what mercenary I put in the the corners they almost always attacked the back two corners. I took out my witch, since they couldn't hit her. I went ahead and switched my mercenaries up and didn't keep my columns together like in the others, but this is what I got:

9.2% / 8.6% / 9.4%

9.0% / 14.6% / 8.8%

15.0% / 9.8% / 15.6%

I haven't been able to test out other formations fully, but it seems that when I switch the columns up and mercenaries, the top priority mercenaries for mobs are Monk, Main, Princess, and Cannoneer(not me, the mercenary). I believe the AI uses a system of values based on what the mercenary is, what is in front of the mercenary, and which row the mercenary is located in. I did some non formation testing, basically only having 3 mercs in the front row and nothing behind them. I would switch out the mercenaries after 20 battles. Some suprisingly noticable things started to show up.

 

I noticed when I moved my monk from the back row to the front, he was attacked 64.1% of the time, instead of any other mercenary (including cloth wearers). This leads me to believe that Monk is the highest priority for any mob. It might be different if you put a staff main up front, but when I put my cannon main upfront, it was attacked 62% of the time. And last but not least, when I put my princess upfront she was attacked 50.4% of the time. Lady Knight and Spartan seem to be the least prefered with 26.2% and 29% respectively.

 

Hopefully this will help anyone who read through all that. Basically put your Monk and Main behind some strong defenders (Lady Knight and Spartan perhaps?), and the % even out a little bit.

 

 

===Update 3/08/09===

 

Did some more testing, this time about the priority mobs give certain Mercenaries. In order to determine which mercenary gets attacked more often than the others I only had 3 units in my formation at the time, all in the front row. In such a fashion:

Mercenary 1 / Mercenary 2 / Cannon(main)

I kept Mercenary 1 as the mercenary I was testing. Like if I was testing the priority of the Monk, I put the Monk in the Mercenary 1 slot. After 20 attacks with each mercenary in the game (except witch) in the Mercenary 2 slot, I would switch it to the next mercenary. Also after each attack I rotated the columns, like in my original post. I only tallied the first attack of each battle, this was to make sure that the HP of the merc had nothing to do with the results. All mercenaries were level 120, with full growth vials and skills (including potential), and in +0 level 60 Anu armor and weapons with no rings or charms. I did not count attacks made against my main in the 20 attacks.

 

Now for the results, in order of most attacks against themto the least, out of 400 attacks.

 

1) Monk - 289 attacks

2) Shaman - 272 attacks

3) Elementalist - 271 attacks

4) Dark Archer - 249 attacks

5) Oracle - 247 attacks

6) Janisary - 241 attacks

7) Pirate - 221 attacks

8) Princess - 209 attacks

9) Prophet - 202 attacks

10) Gunner - 196 attacks

11) Inventor - 194

12) Spearman - 193 attacks

13) Artilleryman - 183 attacks

14) Norther Viking - 181 attacks

15) Exorcist - 180 attacks

16) Cannoneer - 173 attacks

17) Beast Trainer - 162 attacks

18) Swordsman - 148 attacks

19) Lady Knight - 124 attacks

20) Spartan - 102 attacks

 

There aren't enough merc rooms to do mercenaries against themselves and I was lazy and did not do a normal Viking or Archer (not leveling them upto level 120 lol). Also Witch could not be done, since normal attacks cannot hit her.

 

Some important things I need to point out that I noticed while doing this. I know alot of people are chosing Shaman over Oracle. It seems however that the Oracle has a much lower priority than a shaman. This repeats itself over and over with other "replacement" mercenaries. Like Spartan and Swordman/Spearman also. It seems as though mobs will chose a Grade D mercenary over a Grade C more often. This may be due to the "upgraded" mercenary having more defense/hp/etc, or perhaps there really is a list in the AI with a formula to determine which mob should attack which mercenary. But that's just speculation hehe. I hope this helps you guys even more when deciding which mercs to place upfront in your builds.

 

My next project is to hopefully find a way to see what effects certain mercenaries have on the priority when behind a certain other merc. Maybe a Spartan infront of a Monk will be less likely to be attacked than an Elementalist behind a Shaman. This one might take me awhile longer, since there is alot of different combinations to do.