Final Fantasy VI

Father and Son Data Party

My son has been playing the old SNES game Final Fantasy VI on iOS. Being one of the truly cherished games from my childhood, we decided to make a fun data project out of it. It's a great way to expose him to data science and programming with a topic he enjoys.

For any RPG one of the best sources of data is its (likely) extensive bestiary. Final Fantasy VI is no exception, upping the enemy count from its predecessors as well as the range of their abilities. The mobile re-releases of the game in the 2010s brought numerous end-game enemy additions for high level players. After searching online we found a pair of text file bestiary guides, one covering only the original SNES enemies and one with the recent mobile version enemy roster. We chose the modern version which would allow us to do some investigation of the newly included baddies.

We did some further work using the Final Fantasy Wiki to label each enemy as a boss or normal encounter. We thought this would allows us to see if we could predict whether or not an enemy was a boss based on their basic character data (see below). In addition to building a basic classification model we also did some exploratory data analysis on the stats for each enemy.

There are various text categories for each enemy as well, such as their location, spells and abilities they're weak against, items to steal or dropped, and their attack animations. These variables could be dummied and investigated in a model, but they also make for a simple dataset ripe for text analysis.

Bestiary Dataset Features

(SNES in parenthesis)

Feature	Description	Var Type	Min	Max
Name	Mobile version name	str	Abaddon	Zurvan
SNES Name	Original SNES name	str	1st Class	Zone Eater
Location	Place encountered	str	Airship	Zozo
Level	Level	int	Yes	99
HP	Hit Points	int	Yes	65,500 (63,000)
MP	Magic Points	int	0	65,000 (60,000)
Attack	Physical Att. Power	int	Yes	250
Magic Attack	Magical Att. Power	int	0	100 (55)
Defense	Physical Def. Power	int	0	255
Magic Defense	Magical Def. Power	int	Yes	255
Evasion	Physical Evasion	int	0	255 (250)
Magic Evasion	Magical Evasion	int	0	255 (250)
Speed	Speed	int	Yes	99
Gil	Gil won by player	int	0	50,000 (30,000)
EXP	Experience won	int	0	50,000 (14,396)
Steal	Items to steal	list	Air Anchor	X-Potion
Drops	Items dropped at death	list	Air Knife	Zwill Crossblade
Elemental Immunity	Elements immune to	list	Earth	Wind
Weak Against	Elements weak vs.	list	Earth	Wind
Absorbs	Elements absorbed	list	Earth	Wind
Type	Special monster type	category	N/A	Undead
Status Immunity	Statuses immune to	list	Berserk	Zombie
Vulnerable To	Statuses weak vs.	list	Berserk	Zombie
Inherent Status	Default statuses	list	Berserk	Zombie
Other Immunity	Ability immunities	list	Control	None
Rage	Attack by Gau's Rage	str	1000 Needles	Wing Snap
Sketch	Attacks by Relm's Sketch	list	Absolute Zero	Yawn
Control	Attack by Relm's Control	str	Acid Rain	Zombie Fang
Metamorphose	Items Metamorphosed into	list	Angel Ring	Viper Darts
MP Kill	Kill with MP attack	bool	False	True
Imp Criticals	Critical hits as Imp	bool	False	True
Run Difficulty	Ease to run from	category	Easy	Can't Run Away
Attack Visual	Animation for each attack	str	Chain Flail	Trident

Can We Predict if an Enemy is, Like, a Boss?

In the interest of keeping things simple we ran an out-of-the-box Random Forest classifier as our model.
To get a more robust picture of the model's learning of the dataset we ran the model fifteen times using a random splitting of the dataset each time (70% train, 30% test) to make our predictions. We aggregated the results and report them below.

Class Balance

The total number of enemies in the bestiary is 406. Of these, 110 are classified as being a, or part of a, boss encounter. This gives a 72:27 ratio of normal-to-boss enemies. This imbalance means that our baseline for any model classifying enemies as boss-or-not is ~72%, since if we blindly labeled every single enemy in the bestiary as "not boss" we'd be right 72% of the time.

Category	Count
Total Enemies	406
Bosses	110

There are a few steps we can take counteract this inherent imbalance; we'll use only two basic ones here, but they're still effective and smart to utilize. The first method was to stratify the data when we split it, ensuring a proportionate distribution of classes in both the training and testing data. The second was to use a weighted scoring metric which mathematically rewards or diminishes the value of each classification based on the class weights.

Prediction Accuracy

Using the weighted F1 Score, we were able to achieve appreciable success. The average training data F1 score was around 94.55% with the average testing data F1 score slightly below at 92.13%, as expected. Pruning of the trees in the Random Forest helped reduce overfitting to narrow the gap between the two. (If there's a wide gap between your training and testing scores, you're likely overfitting your model). Also as expected, the variance in the training scores for all fifteen predictions was smaller (0.0083) than that for the testing scores (0.0220).

15 Random Runs with All Enemies

Run	Train F1 (%)	Test F1 (%)	∆
1	94.17	89.75	4.42
2	93.50	92.83	0.67
3	95.85	92.72	3.13
4	94.50	88.99	5.51
5	93.15	94.29	-1.14
6	94.18	88.99	5.19
7	93.17	92.10	1.07
8	95.19	95.92	-0.73
9	94.85	89.48	5.37
10	95.85	93.60	2.25
11	95.50	92.83	2.67
12	94.17	90.57	3.60
13	94.50	91.28	3.22
14	94.45	95.96	-1.51
15	95.13	92.59	2.54
Avg.	94.55	92.13	2.42
Std.	0.0083	0.0220	-.0137

Feature Importance

The most predictive feature in the dataset for determining whether an enemy was a boss or not was (*drumroll emoji*)... Experience Won and Gil Won!

Feature Ranks

Rank	Feature	Importance
1	Experience	31.51%
2	Gil	22.24%
3	MP	14.38%
4	HP	10.90%
5	Level	5.29%
6	Magic_Attack	5.25%
7	Speed	2.58%
8	Attack	1.98%
9	Defense	1.36%
10	Evasion	1.34%
11	Dragon_Den	1.02%
12	Magic_Defense	0.93%
13	iOS	0.72%
14	Magic_Evasion	0.49%

This is somewhat surprising initially because our first hunch was that the characteristics of the bosses, such as their magic attack or physical defense or evasion would separate them from normal enemies. But this is where understanding the data comes into play.

We must remember that feature importance doesn't tell us whether having more or less of something is important for predicting the target, only that having one of those two characteristics -- high or low in some variable -- is important. A quick glance at the mean values for the experience that bosses give the player when defeated shows the reason why: bosses don't give Experience or Gil!

That's an unexpected finding. Generally speaking, bosses are the toughest enemies in the game at the point in which they're encountered. You'd expect to get the most experience and the most money upon victory. In many RPGs this is the case. In Final Fantasy VI (and, perhaps, in other early-generation Final Fantasies) it isn't. The reward for beating most bosses is, apparently, continuation of the story and a bit of personal satisfaction. Actually, we all know the real purpose of boss fights is getting to hear the decisively awesome boss battle theme. Awwwww yeahhhhh!

There is a distinction to make here, however. As mentioned earlier the mobile versions of the game released in the last few years brought about the addition of end-game bosses (and an end-game dungeon). These bosses do give experience. Considering they have no relevance to the story, which is originally what defined a boss battle -- an enemy the player must defeat in order to advance the game's story -- and that they weren't originally included in the game, I think a fair investigation can exclude them from the analysis.

Doing so gives a smaller dataset of only 366 enemies with 96 bosses, and ups the predictive ability of the model since only one of the bosses of those 96 gives experience.

SNES Enemies Only Dataset

Category	Count	Num. Giving Exp	% Giving Exp	Num. Giving Gil	% Giving Gil
Total Enemies	366	246	67.2%	249	68.0%
Non-Bosses	270	245	90.7%	238	88.1%
Bosses	96	1	1.04%	11	11.4%

With this trimmed, SNES-only dataset we can see why the model latches onto Experience as the biggest predictor for whether an enemy is a boss or not. Over 90% of non-bosses give the player some experience upon being defeated, while only 1% of bosses do (side note: the only boss to give experience is the "Hell's Rider" enemy, the mounted soldier patrolling the wintry peaks above Narshe during Kefka's World of Balance assault). The same goes for Gil as well, with bosses giving none.

The feature importances remain the same as well, with slightly increased Experience importance and zero importance for Dragon Den.

The classification results improve as well, showing that the SNES-only database is slightly easier to make predictions on because of the lack of experience-giving bosses. With a mean test F1 score of 96.87% from fifteen randomized runs, it's clear we can detect whether an enemy would be a boss or not in the original SNES version of Final Fantasy VI.

SNES Enemies Only Mean Predictions

Here are the results from running the model fifteen times with only the SNES enemies list.

Measure	Train F1	Test F1	∆
Mean	98.14%	96.87%	1.27
Std.	.0047	.0124	-.0076

Visual Investigation

With the results from our model in hand, it is usually fun to do some visual inspection of the features or trends reported above.

Experience as Level Increases

With bosses colored red and non-bosses blue, we see that bosses just don't give experience regardless of how high their level is. As non-bosses become higher in level they give more experience. This plot confirms what our model picked up, as shown in their trend line that climbs rapidly. Bosses would have a flat trend line, of course.

Bosses Have More HP and MP

Referencing the feature importances above, HP and MP are ranked 3rd and 4th, respectively. This makes sense as we'd expect bosses to have more of both statistics since they are meant to be more challenging enemies. Their values in these two stats just happens to vary some, which makes sense, while Experience does not, giving it the stronger predictive power. The mean for the SNES-only population of enemies is marked by dashed lines for both axes. We see an indistinguishable cluster of non-bosses (blue) in the bottom left of the plot, while bosses are distributed at a much higher range of both HP and MP. We will quantify this below.

Bosses Have More HP per Level

To belabor the point, bosses have more health per level as we'd expect from the previous plot. The same is true for MP.

Bosses Are Average in Other Attributes

We'd think bosses would have higher values in all stats than normal enemies, but this isn't the case. It's only true for HP and MP. Here we look at their Defense and Magic Defense and see no separation at all from non-bosses.

The Probability of Being a Boss

For each enemy in the database the model makes a prediction: is this enemy a boss or not? While the final classification is binary (yes or no), the degree to which the model "believes" in this classification is not. For each enemy the model assigns a probability that they are indeed what the model has classified them as; in other words, the model tells us how confident it is about each enemy being a boss or not.

I thought it would be fun to look at these probabilities to see exactly which enemies the model says are the most "boss-like". Here's a table of the fifteen most probable bosses according to this model (again, after aggregating fifteen randomly selected runs), with a full list in the appendix at the end of this post.

Most Probable Bosses

Rank	SNES_Name	Actual Boss?	Probability	Prediction	Correct?	Location
1	White Drgn	Yes	95.64%	Boss	Yes	Cultists' Tower
2	Doom	Yes	95.62%	Boss	Yes	Kefka's Tower (Final Battle)
3	Blue Drgn	Yes	95.32%	Boss	Yes	Ancient Castle
4	Sleep	Yes	95.30%	Boss	Yes	Kefka's Tower (Final Battle)
5	Master Pug	Yes	95.28%	Boss	Yes	Cave to Ancient Castle
6	Red Dragon	Yes	95.28%	Boss	Yes	Phoenix Cave
7	Storm Drgn	Yes	95.26%	Boss	Yes	Mt. Zozo
8	Tiger	Yes	95.21%	Boss	Yes	Kefka's Tower (Final Battle)
9	Ice Dragon	Yes	95.15%	Boss	Yes	Narshe Cliffs
10	Gold Drgn	Yes	95.04%	Boss	Yes	Kefka's Tower
11	Tools	Yes	94.85%	Boss	Yes	Kefka's Tower (Final Battle)
12	Inferno	Yes	94.80%	Boss	Yes	Kefka's Tower
13	Short Arm	Yes	94.61%	Boss	Yes	Kefka's Tower (Final Battle)
14	Striker	Yes	94.57%	Boss	Yes	Kefka's Tower
15	Face	Yes	94.40%	Boss	Yes	Kefka's Tower (Final Battle)

For fun let's look at the least likely bosses.

Least Probable Bosses

Rank	SNES_Name	Actual Boss?	Probability	Prediction	Correct?	Location
1	Vindr	No	3.06%	Normal	Yes	Owzer's Mansion (WoR)
2	Ralph	No	3.18%	Normal	Yes	Grass Areas (WoR)
3	Muus	No	3.18%	Normal	Yes	Grass Areas (WoR)
4	Maliga	No	3.19%	Normal	Yes	Figaro Continent (WoR)
5	Exoray	No	3.20%	Normal	Yes	Darril's Tomb (WoR)
6	Wild Cat	No	3.22%	Normal	Yes	Owzer's Mansion (WoR)
7	Poppers	No	3.22%	Normal	Yes	Umaro's Cave
8	Iron Fist	No	3.23%	Normal	Yes	Kohlingen Forest
9	Osprey	No	3.23%	Normal	Yes	Southern Continent (WoR)
10	Gigan Toad	No	3.24%	Normal	Yes	Southern Continent (WoR)

These results fit well with common sense. Many of the top fifteen probable bosses are dragons or are found in Kefka's Tower, the game's final dungeon. Indeed six of the fifteen most probable are from the final battle itself. Conversely the least likely to be bosses are nearly all random Overworld enemies.

Hypothesis Testing

Hypothesis testing is one way we can determine whether or not two groups or classes -- such as bosses vs. non-bosses -- are mathematically different from each other (in whatever metric we choose). Once analyzed, if they are statistically separated beyond a given threshold, we can say they independent from each other with an certain level of confidence (chosen by us; higher confidence means more difficult threshold to pass). If they are not different enough we can't say that they are actually the same. Instead all we can say is that the two groups are not different enough to rule out that their measurements merely occurred by chance (again, we determine how strict "chance" is here).

As an example if we measured the difference between bosses' Experience given and non-bosses, we'd find that yes, there is clear statistical evidence that they are different groups (one being basically zero, the other in the hundreds or thousands).

Means to an End

(Bold if at least double the other)

Attribute	Non-Bosses Mean	Bosses Mean	Larger	X-times Larger
Exp	1,029.4	4.1	Non	251.0x
Gil (No KatanaSoul) *	577.1	53.8	Non	10.7x
HP	2,775.7	19,841	Boss	7.1x
MP	1,392.2	7,101.6	Boss	5.1x
Evasion	10.4	5.8	Non	1.8x
Level	28.6	42.1	Boss	1.4x
Attack	15.6	23	Boss	1.4x
Speed	34.2	42.8	Boss	1.2x
Magic_Attack	10.3	8.8	Non	1.1x
Magic_Evasion	1.85	1.6	Non	1.1x
Defense	112.0	116.4	Boss	1.0x
Magic_Defense	139.3	143.6	Boss	1.0x
Sample Size	272	96	Non	(2.8x)

* Only seven bosses give over 100 gil. Only one boss, KatanaSoul, gives appreciable gil. In fact, he gives 30,000 Gil which not only skews the entire category for bosses but is also the single largest purse rewarded by any enemy in the entire SNES game (second highest in the mobile versions). Taking him out, the mean gil given by bosses becomes a paltry 53 gil which is far more representative of the other 95 Bosses.

This breakdown of the means per category between the two classes of enemies gives us great insight into what we could investigate with a hypothesis test. It is not surprising to see the categories of the four largest percentage differences are also the four features the model identified as being most important in predicting whether an enemy is a boss or not. Thus the four features we'd like to test are the four largely different ones, and Evasion because it might be close to being significant as well.

The basic Z-test and t-test make the assumption that our population is normally distributed -- after all, that's what the Z-score and t-score mean: how far our observed sample mean is from the population's/other sample's mean in a normal distribution. So, we need to test to see if our data is normally distributed.

To do this we can use a QQ plot to analyze the residuals of our data. Keeping it simple, we want to see our data follow the straight line which would match the theoretical distribution (i.e. normal distribution in this case) our of data. Since HP and MP are quite similar, I will only use HP.

QQ Plots for Normality

Well, that's not a good result. Of the three features that were important for the model, none are normally distributed. The QQ plot is just a visual check but is very instructive in this case. To be fair, we saw above in the scatter plots and % Giving table that Exp and Gil are very much not normally distributed. The QQ plots only confirm this. Defense is also the feature which is "most normal." If the other features were also as close to normality we could do further tests like the Shapiro-Wilk analysis to try to quantify "how normal" they were. But we don't even need to do that as they very clearly are not.

We can still do the Z-test (since we do have over 30 samples and know the standard deviations), but its meaning is left uncertain. If the results say that the mean for Exp for non-bosses is 10σ greater than the mean for bosses, what meaning does that really have when the underlying distribution itself isn't even remotely close to normal? Pushing it farther is beyond the scope of this write-up, so instead we will simply carry out the tests per normal just to get a sense of these categories between bosses and non-bosses for demonstrational purposes.

Z-Scores

Category	Boss Z Score (σ)
Exp.	-10.04
Gil	-7.97
HP	9.08
MP	4.75
Defense	-1.59

Pretending these features did come from roughly normal distributions, we'd be able to reject the null hypothesis that bosses and non-bosses have indistinguishable statistics in the four most "important" categories according to our model. Again, intuitively this fits well -- the model deemed these as the most predictive features so it is sensible to see that there are real distinctions between bosses and non-bosses in them. Defense, the one variable whose roughly Gaussian distribution makes this hypothesis test valid, just misses passing our threshold of the 95-percent confidence level (-1.65), meaning we could not say statistically that a boss' defense was different than a non-bosses beyond chance.

To-Do

1. Parse text features like Location, Steal, or Drop to get the items, rates, etc. for each and dummy those for more insight. I bet immunities scale well with bosses....

Appendix

SNES_Name	Actual Boss	Probability	Prediction	Right Prediction?
White Drgn	Yes	0.956485	Boss	Yes
Doom	Yes	0.956232	Boss	Yes
Blue Drgn	Yes	0.953287	Boss	Yes
Sleep	Yes	0.953098	Boss	Yes
Master Pug	Yes	0.952831	Boss	Yes
Red Dragon	Yes	0.952810	Boss	Yes
Storm Drgn	Yes	0.952680	Boss	Yes
Tiger	Yes	0.952158	Boss	Yes
Ice Dragon	Yes	0.951512	Boss	Yes
Gold Drgn	Yes	0.950414	Boss	Yes
Tools	Yes	0.948505	Boss	Yes
Inferno	Yes	0.948053	Boss	Yes
Short Arm	Yes	0.946134	Boss	Yes
Striker	Yes	0.945745	Boss	Yes
Face	Yes	0.944032	Boss	Yes
Phunbaba #3	Yes	0.943571	Boss	Yes
Dullahan	Yes	0.942945	Boss	Yes
Phunbaba #4	Yes	0.942805	Boss	Yes
Skull Drgn	Yes	0.942409	Boss	Yes
Phunbaba	Yes	0.941779	Boss	Yes
Ultros #3	Yes	0.940362	Boss	Yes
Doom Gaze	Yes	0.940050	Boss	Yes
Hit	Yes	0.939855	Boss	Yes
Magic	Yes	0.936566	Boss	Yes
Atma	Yes	0.933805	Boss	Yes
Hidon	Yes	0.932706	Boss	Yes
Kefka	Yes	0.932470	Boss	Yes
Phunbaba #2	Yes	0.931509	Boss	Yes
Wrexsoul	Yes	0.930727	Boss	Yes
Chadarnook	Yes	0.929414	Boss	Yes
Chadarnook #2	Yes	0.929391	Boss	Yes
Umaro	Yes	0.927943	Boss	Yes
AtmaWeapon	Yes	0.927535	Boss	Yes
Poltrgeist	Yes	0.927480	Boss	Yes
SrBehemoth #2	Yes	0.926387	Boss	Yes
SrBehemoth	Yes	0.920595	Boss	Yes
Girl	Yes	0.920220	Boss	Yes
Tritoch #3	Yes	0.919888	Boss	Yes
Number 128	Yes	0.919850	Boss	Yes
Moe	Yes	0.919292	Boss	Yes
Dirt Drgn	Yes	0.918232	Boss	Yes
Goddess	Yes	0.917442	Boss	Yes
FlameEater	Yes	0.916002	Boss	Yes
Tentacle #2	Yes	0.915898	Boss	Yes
Larry	Yes	0.915078	Boss	Yes
Tentacle	Yes	0.914851	Boss	Yes
Curley	Yes	0.913987	Boss	Yes
Tritoch #2	Yes	0.913260	Boss	Yes
SoulSaver	Yes	0.909669	Boss	Yes
Tentacle #4	Yes	0.908459	Boss	Yes
MissileBay	Yes	0.905828	Boss	Yes
Tritoch	Yes	0.905649	Boss	Yes
Tentacle #3	Yes	0.904011	Boss	Yes
Chupon #2	Yes	0.900924	Boss	Yes
Chupon	Yes	0.900552	Boss	Yes
Guardian	Yes	0.899279	Boss	Yes
Ultros	Yes	0.898088	Boss	Yes
Ultros #2	Yes	0.896964	Boss	Yes
Rough	Yes	0.894461	Boss	Yes
Ultros #4	Yes	0.894097	Boss	Yes
Crane #2	Yes	0.892818	Boss	Yes
Number 024	Yes	0.891203	Boss	Yes
Ifrit	Yes	0.888766	Boss	Yes
Kefka #2	Yes	0.884097	Boss	Yes
Long Arm	Yes	0.881370	Boss	Yes
Hidonite	Yes	0.878668	Boss	Yes
Crane	Yes	0.877504	Boss	Yes
Hidonite #4	Yes	0.874953	Boss	Yes
Hidonite #2	Yes	0.874386	Boss	Yes
Guardian #2	Yes	0.872717	Boss	Yes
Hidonite #3	Yes	0.868432	Boss	Yes
Shiva	Yes	0.867825	Boss	Yes
Left Blade	Yes	0.862286	Boss	Yes
Siegfried #2	Yes	0.861657	Boss	Yes
Pugs	Yes	0.845212	Boss	Yes
Laser Gun	Yes	0.838951	Boss	Yes
Head	Yes	0.834829	Boss	Yes
Air Force	Yes	0.826384	Boss	Yes
Specter	0	0.824626	Boss	No
KatanaSoul	Yes	0.810851	Boss	Yes
Whelk	Yes	0.809127	Boss	Yes
GhostTrain	Yes	0.802074	Boss	Yes
Intangir	0	0.801050	Boss	No
Vargas	Yes	0.788006	Boss	Yes
Whelk Head	Yes	0.774232	Boss	Yes
MagiMaster	Yes	0.756835	Boss	Yes
M-TekArmor	Yes	0.750494	Boss	Yes
Nerapa	Yes	0.737887	Boss	Yes
Presenter	Yes	0.736911	Boss	Yes
RightBlade	Yes	0.736239	Boss	Yes
Speck	Yes	0.724175	Boss	Yes
Rizopas	Yes	0.699169	Boss	Yes
Allo Ver	0	0.684843	Boss	No
Dadaluma	Yes	0.679066	Boss	Yes
Ipooh	Yes	0.639694	Boss	Yes
Naughty	0	0.633450	Boss	No
Telstar	0	0.618884	Boss	No
Marshal	Yes	0.568641	Boss	Yes
Siegfried	Yes	0.567093	Boss	Yes
TunnelArmr	Yes	0.566835	Boss	Yes
Piranha	0	0.482761	No	Yes
Soldier #2	0	0.423284	No	Yes
Brachosaur	0	0.414558	No	Yes
Land Worm	0	0.411210	No	Yes
L.90 Magic	0	0.404199	No	Yes
Nohrabbit	0	0.377637	No	Yes
Scullion	0	0.358531	No	Yes
L.60 Magic	0	0.340282	No	Yes
L.80 Magic	0	0.340156	No	Yes
L.50 Magic	0	0.327576	No	Yes
L.70 Magic	0	0.316479	No	Yes
Magic Urn	0	0.299061	No	Yes
Dragon	0	0.298808	No	Yes
Gigantos	0	0.297205	No	Yes
Tyranosaur	0	0.297165	No	Yes
L.30 Magic	0	0.291166	No	Yes
L.40 Magic	0	0.289285	No	Yes
Cadet	0	0.283220	No	Yes
L.20 Magic	0	0.283132	No	Yes
L.10 Magic	0	0.274864	No	Yes
Behemoth	0	0.269365	No	Yes
Doberman	0	0.269022	No	Yes
Didalos	0	0.268025	No	Yes
Leader	0	0.259767	No	Yes
Rider	Yes	0.251856	No	No
EarthGuard	0	0.248088	No	Yes
Hoover	0	0.244752	No	Yes
Veteran	0	0.243489	No	Yes
Templar	0	0.240711	No	Yes
Soldier	0	0.235018	No	Yes
Cactrot	0	0.229134	No	Yes
Mover	0	0.221985	No	Yes
Prometheus	0	0.217618	No	Yes
Mega Armor	0	0.214704	No	Yes
Grunt	0	0.212900	No	Yes
Doom Drgn	0	0.210766	No	Yes
Peepers	0	0.207570	No	Yes
Sp Forces	0	0.206382	No	Yes
Zone Eater	0	0.192845	No	Yes
Vomammoth	0	0.176938	No	Yes
Steroidite	0	0.176694	No	Yes
Brontaur	0	0.176499	No	Yes
GtBehemoth	0	0.170180	No	Yes
HeavyArmor	0	0.163533	No	Yes
Mantodea	0	0.157745	No	Yes
Dark Force	0	0.155472	No	Yes
Fortis	0	0.152340	No	Yes
Lethal Wpn	0	0.147822	No	Yes
Trixter	0	0.145940	No	Yes
Necromancr	0	0.141195	No	Yes
Outsider	0	0.140026	No	Yes
Goblin	0	0.138056	No	Yes
Retainer	0	0.136109	No	Yes
Dahling	0	0.134894	No	Yes
Crawler	0	0.134885	No	Yes
Grenade	0	0.131588	No	Yes
Evil Oscar	0	0.129909	No	Yes
Vectagoyle	0	0.126831	No	Yes
Sky Base	0	0.126471	No	Yes
Chaos Drgn	0	0.123935	No	Yes
Hemophyte	0	0.119457	No	Yes
Aspik	0	0.118851	No	Yes
Pug	0	0.114497	No	Yes
Mad Oscar	0	0.108689	No	Yes
Rhyos	0	0.107101	No	Yes
Displayer	0	0.105119	No	Yes
Phase	0	0.105027	No	Yes
Brainpan	0	0.102843	No	Yes
Madam	0	0.102532	No	Yes
Black Drgn	0	0.101233	No	Yes
Geckorex	0	0.097787	No	Yes
Dueller	0	0.096390	No	Yes
Opinicus	0	0.095093	No	Yes
FossilFang	0	0.094904	No	Yes
Allosaurus	0	0.094897	No	Yes
Warlock	0	0.093887	No	Yes
Borras	0	0.093219	No	Yes
Aquila	0	0.089355	No	Yes
TumbleWeed	0	0.086053	No	Yes
Samurai	0	0.084244	No	Yes
Sprinter	0	0.083980	No	Yes
Crusher	0	0.083612	No	Yes
Uroburos	0	0.083260	No	Yes
Lobo	0	0.083017	No	Yes
Spek Tor	0	0.081628	No	Yes
Still Life	0	0.080982	No	Yes
IronHitman	0	0.079793	No	Yes
HadesGigas	0	0.079664	No	Yes
Parasoul	0	0.078744	No	Yes
Reach Frog	0	0.076804	No	Yes
Boxed Set	0	0.076108	No	Yes
Pterodon	0	0.076019	No	Yes
ChickenLip	0	0.075282	No	Yes
Nightshade	0	0.074196	No	Yes
Chaser	0	0.073190	No	Yes
Sky Cap	0	0.072534	No	Yes
Over-Mind	0	0.071715	No	Yes
Harvester	0	0.071584	No	Yes
Ninja	0	0.071427	No	Yes
Vectaur	0	0.071297	No	Yes
Parasite	0	0.070652	No	Yes
Spit Fire	0	0.070560	No	Yes
Mag Roader #2	0	0.069588	No	Yes
Test Rider	0	0.069105	No	Yes
PlutoArmor	0	0.068910	No	Yes
Bomb	0	0.068097	No	Yes
Ghost	0	0.068014	No	Yes
Anguiform	0	0.068003	No	Yes
Sky Armor	0	0.067638	No	Yes
Chimera	0	0.066827	No	Yes
Orog	0	0.065812	No	Yes
Pm Stalker	0	0.063962	No	Yes
Guard	0	0.062936	No	Yes
Spectre	0	0.062709	No	Yes
HermitCrab	0	0.062582	No	Yes
ProtoArmor	0	0.062444	No	Yes
Mag Roader #3	0	0.061996	No	Yes
Scorpion	0	0.061475	No	Yes
Cruller	0	0.060974	No	Yes
Osteosaur	0	0.059527	No	Yes
Hazer	0	0.059419	No	Yes
Wirey Drgn	0	0.059264	No	Yes
PowerDemon	0	0.059185	No	Yes
Were-Rat	0	0.059065	No	Yes
Bloompire	0	0.059044	No	Yes
Gobbler	0	0.058531	No	Yes
Zombone	0	0.057317	No	Yes
Enuo	0	0.056991	No	Yes
Latimeria	0	0.056326	No	Yes
Sand Horse	0	0.056161	No	Yes
Innoc	0	0.055614	No	Yes
Dark Side	0	0.055513	No	Yes
Woolly	0	0.054749	No	Yes
Covert	0	0.054392	No	Yes
Tap Dancer	0	0.054212	No	Yes
Wart Puck	0	0.054150	No	Yes
Figaliz	0	0.051151	No	Yes
Adamanchyt	0	0.049496	No	Yes
Toe Cutter	0	0.049348	No	Yes
Pan Dora	0	0.048148	No	Yes
Ogor	0	0.047425	No	Yes
Harpy	0	0.046430	No	Yes
Dante	0	0.046291	No	Yes
Rhinotaur	0	0.045741	No	Yes
Prussian	0	0.045635	No	Yes
Karkass	0	0.045020	No	Yes
Critic	0	0.044856	No	Yes
Repo Man	0	0.044146	No	Yes
Rain Man	0	0.044016	No	Yes
SoulDancer	0	0.043891	No	Yes
Barb-e	0	0.042255	No	Yes
Tusker	0	0.042229	No	Yes
Sea Flower	0	0.042116	No	Yes
Sand Ray	0	0.041879	No	Yes
Punisher	0	0.041182	No	Yes
Vaporite	0	0.040921	No	Yes
Suriander	0	0.040680	No	Yes
Misfit	0	0.040589	No	Yes
NeckHunter	0	0.040583	No	Yes
B.Day Suit	0	0.040286	No	Yes
Bleary	0	0.040172	No	Yes
Crawly	0	0.039902	No	Yes
Leafer	0	0.039794	No	Yes
GloomShell	0	0.039641	No	Yes
Areneid	0	0.039485	No	Yes
Hornet	0	0.039323	No	Yes
Fidor	0	0.038872	No	Yes
Vermin	0	0.038824	No	Yes
Merchant	0	0.038788	No	Yes
Slatter	0	0.038727	No	Yes
Dark Wind	0	0.038556	No	Yes
GreaseMonk	0	0.038168	No	Yes
Gold Bear	0	0.037579	No	Yes
Buffalax	0	0.037375	No	Yes
Rhodox	0	0.037348	No	Yes
Ursus	0	0.037269	No	Yes
Junk	0	0.036831	No	Yes
Exocite	0	0.036805	No	Yes
Balloon	0	0.036685	No	Yes
Luridan	0	0.036630	No	Yes
Nautiloid	0	0.036600	No	Yes
Apokryphos	0	0.036367	No	Yes
CrassHoppr	0	0.035995	No	Yes
Officer	0	0.035746	No	Yes
Lich	0	0.035712	No	Yes
Cluck	0	0.035675	No	Yes
Actaneon	0	0.035647	No	Yes
Wizard	0	0.035519	No	Yes
Vector Pup	0	0.035478	No	Yes
Trilium	0	0.035296	No	Yes
Stray Cat	0	0.035245	No	Yes
Ing	0	0.035186	No	Yes
Red Fang	0	0.035175	No	Yes
Deep Eye	0	0.035053	No	Yes
1st Class	0	0.035032	No	Yes
Bogy	0	0.035026	No	Yes
Poplium	0	0.034955	No	Yes
Scrapper	0	0.034948	No	Yes
Beakor	0	0.034876	No	Yes
Coelecite	0	0.034795	No	Yes
Harpiai	0	0.034790	No	Yes
Over Grunk	0	0.034785	No	Yes
Eland	0	0.034767	No	Yes
Hipocampus	0	0.034668	No	Yes
Wild Rat	0	0.034617	No	Yes
Flan	0	0.034463	No	Yes
Brawler	0	0.034407	No	Yes
Commando	0	0.034361	No	Yes
Rhobite	0	0.034350	No	Yes
Primordite	0	0.034315	No	Yes
Pipsqueak	0	0.034295	No	Yes
Trooper	0	0.034293	No	Yes
Rinn	0	0.034253	No	Yes
Bug	0	0.034186	No	Yes
Vulture	0	0.034129	No	Yes
Commander	0	0.034110	No	Yes
Mind Candy	0	0.034093	No	Yes
Iron Fist	0	0.033971	No	Yes
Cirpius	0	0.033947	No	Yes
Joker	0	0.033837	No	Yes
Insecare	0	0.033769	No	Yes
Osprey	0	0.033757	No	Yes
Lizard	0	0.033739	No	Yes
Whisper	0	0.033726	No	Yes
Wyvern	0	0.033643	No	Yes
Bounty Man	0	0.033454	No	Yes
Apparite	0	0.033447	No	Yes
Ralph	0	0.033380	No	Yes
StillGoing	0	0.033345	No	Yes
Mag Roader #4	0	0.033289	No	Yes
Slurm	0	0.033252	No	Yes
Gabbldegak	0	0.033203	No	Yes
Abolisher	0	0.033131	No	Yes
Trapper	0	0.033092	No	Yes
Garm	0	0.033058	No	Yes
Kiwok	0	0.033018	No	Yes
Sewer Rat	0	0.033017	No	Yes
Ceritops	0	0.032941	No	Yes
Mandrake	0	0.032885	No	Yes
Rhinox	0	0.032847	No	Yes
Cephaler	0	0.032830	No	Yes
Mesosaur	0	0.032787	No	Yes
Baskervor	0	0.032756	No	Yes
WeedFeeder	0	0.032740	No	Yes
General	0	0.032627	No	Yes
Lunaris	0	0.032595	No	Yes
Delta Bug	0	0.032422	No	Yes
Chitonid	0	0.032390	No	Yes
Mag Roader	0	0.032380	No	Yes
Poppers	0	0.032245	No	Yes
Nastidon	0	0.032228	No	Yes
Tomb Thumb	0	0.032145	No	Yes
Red Wolf	0	0.032122	No	Yes
Trilobiter	0	0.032074	No	Yes
Wild Cat	0	0.032006	No	Yes
SlamDancer	0	0.031907	No	Yes
Drop	0	0.031773	No	Yes
Humpty	0	0.031675	No	Yes
Gilomantis	0	0.031628	No	Yes
Anemone	0	0.031534	No	Yes
Psychot	0	0.031447	No	Yes
Muus	0	0.031393	No	Yes
Maliga	0	0.031201	No	Yes
Vindr	0	0.030892	No	Yes
Gigan Toad	0	0.030774	No	Yes
Exoray	0	0.030213	No	Yes