svm.py

# coding: utf-8

# ## COPY THIS FOR TRAINING DATA

# In[113]:

from sklearn.model_selection import train_test_split
import charba.api as capi
import pandas as pd
import numpy as np
from nba_py.team import *
from nba_py.game import *
from nba_py.player import *
np.set_printoptions(suppress=True)
pd.set_option('display.float_format', lambda x: '%.3f' % x)

from underscore import _
import qgrid
qgrid.nbinstall(overwrite=True)
                


# In[114]:

# GET ALL THE GAMES 
season_game_history = _.flatten([ capi.get_game_history(TEAMS[key]['id']) for key in TEAMS.keys()  ])



# In[115]:

# GET ALL TEAM STATS FOR EVERY GAME PLAYED UP UNTIL NOW 

all_games = pd.DataFrame(_.flatten([x.team_stats() for x in season_game_history]))
all_games = all_games[0]


print(df.loc[0].tolist())

def createWinnerY(df):
    y = 0
    
    if df.loc[0]['PTS'] > df.loc[1]['PTS']:
        y = 1 # df.loc[0]['GAME_ID']
    else: 
        y = 2 # df.loc[1]['GAME_ID']
    return y


# In[117]:

# CREATE TRAIN AND TEST DATA
from underscore import _

print(type(_))

all_X = []
all_y = []
for df in all_games:
    y = createWinnerY(df)
    # Drop all the textual elements and leave  only usable numbers
    if df.columns[0] == 'GAME_ID':
        df.drop(df.columns[[0, 1, 2, 3, 4, 5]], axis=1, inplace=True)
    
    all_X.append(df.values.flatten())
    all_y.append(y)

    
X_train, X_test, y_train, y_test = train_test_split(all_X, all_y, test_size=0.33, random_state=42)



# ## NOW FOR THE SVM 

# In[118]:

from sklearn.svm import SVC
# NOW THE SVM STUFF

print(np.shape(all_games))
print(np.shape(all_X))
print(np.shape(X_train))
print(len(y_train))
#print(np.shape(y_train))
X = np.array(X_train)
y = y_train 
#y = np.array(y_train)

clf = SVC()
clf.fit(X,y)


# ## PREDICTIONS GO HERE
# 

# In[125]:


print("ACCURACY FROM TEST SET : " + str(clf.score(X_test,y_test) ))
predictions = clf.predict(X_test)
predictions

X_test[0]
all_X