-
Notifications
You must be signed in to change notification settings - Fork 0
/
grouped_batters.py
330 lines (301 loc) · 12 KB
/
grouped_batters.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
import numpy as np
import json
from sklearn.linear_model import SGDRegressor
from sklearn.preprocessing import StandardScaler
from sklearn.naive_bayes import GaussianNB
from sklearn.cluster import KMeans
from sklearn import svm
from pymongo import MongoClient
from bson.objectid import ObjectId
from collections import Counter
from sklearn.feature_extraction import DictVectorizer
import matplotlib.pyplot as plt
import pickle
events = {
'Strikeout': 1,
'Strikeout - DP': 1,
'Pop Out': .8,
'Flyout': .7,
'Groundout': .7,
'Forceout': .7,
'Grounded Into DP': .7,
'Double Play': .7,
'Triple Play': .7,
'Fielders Choice': .7,
'Fielders Choice Out': .7,
'Field Error': .7,
'Lineout': .2,
'Runner Out': .1,
'Sac Fly': -.3,
'Sac Fly DP': -.3,
'Hit By Pitch': -1,
'Walk': -1,
'Single': -1,
'Double': -1.3,
'Triple': -1.5,
'Home Run': -2,
'Sac Bunt': 0,
'Batter Interference': 0,
'Intent Walk': 0,
'Bunt Pop Out': 0,
'Bunt Groundout': 0,
'Catcher Interference': 0,
'Bunt Lineout': 0,
'Fan interference': 0,
'Sacrifice Bunt DP': 0
}
def getFeatures(pitch, cluster_num):
features = {}
feature_vec = []
features["start_speed"] = pitch["start_speed"]
features["start_speed^2"] = pitch["start_speed"] ** 2
features["px^2"] = pitch["px"] ** 2
if pitch["px"] == 'R':
features["px"] = pitch["px"]
else:
features["px"] = -pitch["px"]
features["pfx_x^2"] = pitch["pfx_x"] ** 2
features["pfx_x"] = pitch["pfx_x"]
features["pfx_z^2"] = pitch["pfx_z"] ** 2
features["pfx_z"] = pitch["pfx_z"]
features["pz-mid"] = pitch["pz"] - (pitch["sz_top"] + pitch["sz_bot"])/2
features["pz-mid^2"] = (pitch["pz"] - (pitch["sz_top"] + pitch["sz_bot"])/2) ** 2
features["break_angle"] = pitch["break_angle"]
features["break_length"] = pitch["break_length"]
features["break_y"] = pitch["break_y"]
features["px*pz-mid"] = features["px"] * features["pz-mid"]
features["px^2*(pz-mid)^2"] = (features["px"] ** 2) * (features["pz-mid"] ** 2)
features["break_length/break_y"] = pitch["break_length"] / pitch["break_y"]
features["break_length*start_speed"] = pitch["break_length"] * pitch["start_speed"]
if pitch["pitch_type"] == "FF":
features["FF"] = 1
else:
features["FF"] = 0
if pitch["pitch_type"] == "CU":
features["CU"] = 1
else:
features["CU"] = 0
if pitch["pitch_type"] == "SL":
features["SL"] = 1
else:
features["SL"] = 0
if pitch["prev_type"] == "FF":
features["prev_FF"] = 1
else:
features["prev_FF"] = 0
if pitch["prev_type"] == "CU":
features["prev_CU"] = 1
else:
features["prev_CU"] = 0
if pitch["prev_type"] == "SL":
features["prev_SL"] = 1
else:
features["prev_SL"] = 0
attribs = ["start_speed", "start_speed^2", "px^2", "px", "pfx_x^2", "pfx_x",
"pfx_z^2", "pfx_z", "pz-mid", "pz-mid^2", "break_angle", "break_length", "break_y",
"px*pz-mid", "px^2*(pz-mid)^2", "break_length/break_y", "break_length*start_speed"]
for attrib in attribs:
for pitch_type in ["prev_FF", "prev_CU", "prev_SL", "FF", "CU", "SL"]:
features[pitch_type + "_" + attrib] = features[pitch_type] * features[attrib]
zones = ['TL', 'TM', 'TR', 'ML', 'MM', 'MR', 'BL', 'BM', 'BR']
for zone in zones:
features['zone_' + zone] = 0
features['prev_zone_' + zone] = 0
current_zone = getZone(pitch['px'], pitch['pz'], pitch['sz_top'], pitch['sz_bot'])
features['zone_' + current_zone] = 1
if pitch['prev_px'] != None:
prev_zone = getZone(pitch['prev_px'], pitch['prev_pz'], pitch['sz_top'], pitch['sz_bot'])
features['prev_zone_' + prev_zone] = 1
for zone1 in zones:
features[zone1 + '_start_speed'] = features['start_speed']
features[zone1 + 'break_length'] = features['break_length']
features[zone1 + '_FF'] = features['FF']
features[zone1 + '_SL'] = features['SL']
features[zone1 + '_CU'] = features['CU']
features[zone1 + '_prev=FF'] = features['prev_FF']
features[zone1 + '_prev=SL'] = features['prev_SL']
features[zone1 + '_prev=CU'] = features['prev_CU']
for zone2 in zones:
features[zone1 + '*prev=' + zone2] = features['zone_' + zone1] * features['prev_zone_' + zone2]
feature_keys = features.keys()
for key in feature_keys:
features['cluster%d_' % cluster_num + key] = features[key]
return features
def getZone(px, pz, sz_top, sz_bot):
if pz - sz_bot < (sz_top - sz_bot)/3.0:
# lower third
if px < -0.236:
return 'TL'
if px < 0.236:
return 'TM'
else:
return 'TR'
elif pz - sz_bot < (sz_top - sz_bot)/3.0 * 2:
#middle third
if px < -0.236:
return 'ML'
if px < 0.236:
return 'MM'
else:
return 'MR'
else:
#top third
if px < -0.236:
return 'BL'
if px < 0.2363:
return 'BM'
else:
return 'BR'
def train(reg, scaler, num_pitches, vec, players):
client = MongoClient('localhost', 27017)
db = client["pitchfx"]
x = []
y = []
num_added = 0
# for pitch in db.pitches.find({"pitch_type":"FF", "batter":{"$in": batters}}, limit=100000):
for pitch in db.pitches.find({"pitch_type":{"$in": ["FF", "SL", "CU"]}, "type":{"$in": ["S", "X"]}}, limit=num_pitches):
cluster = players.get((pitch['batter'], pitch['year']))
if cluster == None:
continue
feature_vec = getFeatures(pitch, cluster)
if pitch["type"] == 'S':
result = 1
if pitch["type"] == 'B':
continue
if pitch["type"] == 'X':
result = events[pitch["event"]]
x.append(feature_vec)
y.append(result)
num_added += 1
print num_added, ' pitches added'
scaler.fit(vec.fit_transform(x).toarray())
reg.fit(scaler.transform(vec.transform(x).toarray()),y)
print 'finished fitting data'
def testTrain(reg, scaler, num_pitches, vec, players):
client = MongoClient('localhost', 27017)
db = client["pitchfx"]
num_added = 0
total_error = 0
total_error_sq = 0
total_maj_err = 0
total_maj_err_sq = 0
# for pitch in db.pitches.find({"pitch_type":"FF", "batter":{"$in": batters}}, limit=100000):
for pitch in db.pitches.find({"pitch_type":{"$in": ["FF", "SL", "CU"]}, "type":{"$in": ["S", "X"]}}, limit=num_pitches):
cluster = players.get((pitch['batter'], pitch['year']))
if cluster == None:
continue
feature_vec = getFeatures(pitch, cluster)
if pitch["type"] == 'S':
result = 1
if pitch["type"] == 'B':
continue
if pitch["type"] == 'X':
result = events[pitch["event"]]
#print reg.predict(scaler.transform(vec.transform(feature_vec).toarray()))[0]
error = reg.predict(scaler.transform(vec.transform(feature_vec).toarray()))[0] - result
#print 'prediction: ', reg.predict(scaler.transform([feature_vec]))[0], ' result: ', result
num_added += 1
total_error += abs(error)
total_error_sq += error ** 2
majority_error = 1-result
total_maj_err += abs(majority_error)
total_maj_err_sq += majority_error ** 2
print num_added, ' pitches tested'
print 'Train Error = ', total_error, ' error per pitch = ', total_error/float(num_added), 'squared error =', total_error_sq/float(num_added)
print 'Train Error (majority algorithm) = ', total_maj_err, ' error per pitch = ', total_maj_err/float(num_added), 'squared error =', total_maj_err_sq/float(num_added)
def test(reg, scaler, num_pitches, vec, players):
client = MongoClient('localhost', 27017)
db = client["pitchfx"]
num_added = 0
total_error = 0
total_error_sq = 0
total_maj_err = 0
total_maj_err_sq = 0
# for pitch in db.pitches.find({"pitch_type":"FF", "batter":{"$in": batters}}, limit=100000, skip=100000):
for pitch in db.pitches.find({"pitch_type":{"$in": ["FF", "SL", "CU"]}, "type":{"$in": ["S", "X"]}}, limit=100000, skip=num_pitches):
cluster = players.get((pitch['batter'], pitch['year']))
if cluster == None:
continue
feature_vec = getFeatures(pitch, cluster)
if pitch["type"] == 'S':
result = 1
if pitch["type"] == 'B':
continue
if pitch["type"] == 'X':
result = events[pitch["event"]]
error = reg.predict(scaler.transform(vec.transform(feature_vec).toarray()))[0] - result
#print 'prediction: ', reg.predict(scaler.transform([feature_vec]))[0], ' result: ', result
num_added += 1
total_error += abs(error)
total_error_sq += error ** 2
majority_error = 1-result
total_maj_err += abs(majority_error)
total_maj_err_sq += majority_error ** 2
print num_added, ' pitches tested'
print 'Test Error = ', total_error, ' error per pitch = ', total_error/float(num_added), 'squared error =', total_error_sq/float(num_added)
print 'Test Error (majority algorithm) = ', total_maj_err, ' error per pitch = ', total_maj_err/float(num_added), 'squared error =', total_maj_err_sq/float(num_added)
def kmeans_features(player, year):
features = {}
features['avg'] = player['avg_%04d' % year]
features['hr'] = player['hr_%04d' % year]
features['slg'] = player['slg_%04d' % year]
features['so%'] = float(player['so_%04d' % year]) / (player['ab_%04d' % year] + player['bb_%04d' % year])
features['bb%'] = float(player['bb_%04d' % year]) / (player['ab_%04d' % year] + player['bb_%04d' % year])
return features
def classifyWithKmeans(num_clusters):
client = MongoClient('localhost', 27017)
db = client["pitchfx"]
x = []
for player in db.players.find():
for year in range(2008, 2016):
if player.get('h_%d' % year) == None or player.get('ab_%d' % year) < 100:
continue
x.append(kmeans_features(player, year))
kmeans = KMeans(init='k-means++', n_clusters=num_clusters, n_init=10, random_state=1000)
vec = DictVectorizer()
scaler = StandardScaler()
scaler.fit(vec.fit_transform(x).toarray())
kmeans.fit(scaler.transform(vec.transform(x).toarray()))
print json.dumps(vec.inverse_transform(scaler.inverse_transform(kmeans.cluster_centers_)), indent=4)
for i in range(0,8):
print 'cluster %d:' % i, list(kmeans.labels_).count(i)
return (kmeans, scaler, vec)
client = MongoClient('localhost', 27017)
db = client["pitchfx"]
(kmeans, kmeans_scaler, kmeans_vec) = classifyWithKmeans(8)
players = {}
x = []
y = []
colors = []
for player in db.players.find():
for year in range(2008, 2016):
if player.get('h_%d' % year) == None or player.get('ab_%d' % year) < 100:
continue
players[(player["player_id"], year)] = kmeans.predict(kmeans_scaler.transform(kmeans_vec.transform(kmeans_features(player, year)).toarray()))[0]
x.append(player["avg_%d" % year])
y.append(player["hr_%d" % year])
colors.append(players[(player["player_id"], year)])
# plt.scatter(x, y, c=colors, alpha=0.5)
# plt.ylabel('Home Runs')
# plt.xlabel('Batting Average')
# plt.show()
print 'finished mapping players'
vec = DictVectorizer()
scaler = StandardScaler()
num_iters = 100
reg = SGDRegressor(loss='squared_loss', n_iter=num_iters, verbose=2, penalty='l2', alpha= 0.001, learning_rate="invscaling", eta0=0.002, power_t=0.4)
num_pitches = 200000
print num_pitches, 'pitches'
print 'training with num iters = ', num_iters
train(reg, scaler, num_pitches, vec, players)
#print reg.coef_
print json.dumps(vec.inverse_transform([reg.coef_]), sort_keys=True, indent=4)
print len(list(reg.coef_)), "total features"
testTrain(reg, scaler, num_pitches, vec, players)
test(reg, scaler, num_pitches, vec, players)
saves = {
'reg': reg,
'scaler': scaler,
'vec': vec,
}
pickle.dump(saves, open('saves.p', 'wb'))