def topk(self, k):
assert self._computed == True, "You haven't computed wordcount"
param = {
"k": str(k),
OperationParam.list_str: self.list_name,
"Type": "cpp"
}
op = Operation("Word#wordcount_topk_py", param, [])
return scheduler.compute_collect(op)
def topk_pagerank(self, k):
assert self._computed == True, "You haven't computed Pagerank"
param = {
"k": str(k),
OperationParam.list_str: self.list_name,
"Type": "cpp"
}
op = Operation("Graph#pagerank_topk_py", param, [])
topk_list = scheduler.compute_collect(op)
return topk_list
def reduce(self, func):
if hasattr(func, '__call__'):
param = {
OperationParam.lambda_str: func,
OperationParam.list_str: self.list_name
}
op = Operation("Functional#reduce_py", param, [self.pending_op])
return compute_collect(op)
else:
return NotImplemented
def train(self, n_iter=10, alpha=0.1):
assert self.loaded
assert isinstance(n_iter, int)
assert isinstance(alpha, float)
param = {
"n_iter": str(n_iter),
"alpha": str(alpha),
OperationParam.list_str: self.list_name,
"Type": "cpp"
}
op = Operation("SVMModel#SVM_train_py", param, [])
param_list = scheduler.compute_collect(op)
self.param = np.array(param_list[:-1])
self.intercept = param_list[-1]
self.loaded = False
self.trained = True
def train(self, n_iter=10, alpha=0.1, is_sparse=1):
assert self.loaded
assert isinstance(n_iter, int)
assert isinstance(alpha, float)
self.pending_op = Operation("LogisticRegressionModel#LogisticR_train_py",
{
"n_iter" : str(n_iter),
"alpha" : str(alpha),
OperationParam.list_str : self.list_name,
"is_sparse": str(is_sparse),
"Type" : "cpp"
},
[Operation("LogisticRegressionModel#LogisticR_init_py", {"Type" : "cpp"})] \
if self.trained else [self.pending_op]
)
print self.pending_op.op_deps
paramlist = scheduler.compute_collect(self.pending_op)
self.param = np.array(paramlist[:-1])
self.intercept = paramlist[-1]
self.trained = True
def print_all(self):
assert self._computed == True, "You haven't computed wordcount"
param = {OperationParam.list_str: self.list_name, "Type": "cpp"}
op = Operation("Word#wordcount_print_py", param, [])
return scheduler.compute_collect(op)
def collect(self):
param = {OperationParam.list_str: self.list_name}
op = Operation("Functional#collect_py", param, [self.pending_op])
return compute_collect(op)