def getTestData(dict,size=None,featureIndices=None,):
test = pd.read_csv('./input/test.csv')
test=test.fillna(0)
# train=train.fillna(train.mean())
test=test.values
if size:
test=test[:size,:]
if featureIndices:
indices=featureIndices
else:
indices=list(range(1,test.shape[1]))#All indices except ID and except response
# print(indices)
changed=utils.categoricalToNumerical(test[:,2],dict)
print("number of categories of column 2 of train and test:",changed[1])
test[:,2]=changed[0]
X= test[:,indices]
ids=test[:,0]
return X,ids
def getTrainData(size=None,featureIndices=None):
train = pd.read_csv('./input/train.csv')
train=train.fillna(0)
# train=train.fillna(train.mean())
train=train.values
if size:
train=train[:size,:]
if featureIndices:
indices=featureIndices
else:
indices=list(range(1,train.shape[1]-1))#All indices except ID and except response
# print(indices)
changed=utils.categoricalToNumerical(train[:,2])
dict=changed[2]
print("number of categories of column 2 of train:",changed[1])
train[:,2]=changed[0]
X, y = train[:,indices] , train[:,-1]
y=list(map(np.int32,y))
return X,y,dict
from sklearn.ensemble import ExtraTreesClassifier
train = pd.read_csv('./input/train.csv')
# print(train.isnull().sum())
train=train.fillna(0)
# print(train.isnull().sum())
train=train.values
# train=train[0:100,:]
# Build a classification task using 3 informative features
indeces=(range(train.shape[1]-1))
indeces=list(set(indeces)-set([0]))#removing ID
# print(indeces)
# print(train.shape[1])
changed=utils.categoricalToNumerical(train[:,2])
print("number of categories of column 2:",changed[1])
train[:,2]=changed[0]
X, y = train[:,indeces] , train[:,-1]
y=list(map(np.int32,y))
# print(X)
numfeatures=len(indeces)
print(type(X))
# X=X[:,indeces]
# Build a forest and compute the feature importances
forest = ExtraTreesClassifier(n_estimators=250,random_state=0)
forest.fit(X, y)
importances = forest.feature_importances_
std = np.std([tree.feature_importances_ for tree in forest.estimators_],
