Merge pull request #25 from CyberAgentAI/feature/load-data-in-c

Feature: Add data path to train method

ffm/__init__.py

@@ -93,8 +93,10 @@ def read_ffm_model(cls, model_path: str) -> "Model":
 
 
 def train(
-    train_data: Dataset,
+    train_data: Optional[Dataset] = None,
+    train_path: Optional[str] = None,
     valid_data: Optional[Dataset] = None,
+    valid_path: Optional[str] = None,
     eta: float = 0.2,
     lam: float = 0.00002,
     nr_iters: int = 15,
@@ -107,17 +109,21 @@ def train(
     random: bool = True,
     nds_rate: float = 1.0,
 ) -> Model:
-    tr = (train_data.data, train_data.labels)
-    iw = train_data.importance_weights
+    tr, iw = None, None
+    if train_data is not None:
+        tr = (train_data.data, train_data.labels)
+        iw = train_data.importance_weights
 
     va, iwv = None, None
     if valid_data is not None:
         va = (valid_data.data, valid_data.labels)
         iwv = valid_data.importance_weights
 
     weights, best_iteration, normalization, best_va_loss = libffm_train(
-        tr,
+        tr=tr,
+        tr_path=train_path,
         va=va,
+        va_path=valid_path,
         iw=iw,
         iwv=iwv,
         eta=eta,

ffm/libffm.pyx

@@ -56,6 +56,7 @@ cdef extern from "ffm.h" namespace "ffm" nogil:
         ffm_float best_va_loss
 
     ffm_model *ffm_train_with_validation(ffm_problem *Tr, ffm_problem *Va, ffm_importance_weights *iws, ffm_importance_weights *iwvs, ffm_parameter param);
+    ffm_problem *ffm_read_problem(char *path);
 
 
 cdef ffm_problem* make_ffm_prob(X, y):
@@ -174,8 +175,10 @@ cdef object _train(
 
 
 def train(
-    tr,
+    tr=None,
+    tr_path=None,
     va=None,
+    va_path=None,
     iw=None,
     iwv=None,
     eta=0.2,
@@ -205,11 +208,18 @@ def train(
     param.nds_rate = nds_rate
 
     cdef:
-        ffm_problem* tr_ptr = make_ffm_prob(tr[0], tr[1])
+        ffm_problem* tr_ptr
         ffm_problem* va_ptr
         ffm_importance_weights *iw_ptr, *iwv_ptr
 
-    if va is not None:
+    if tr_path is not None:
+        tr_ptr = ffm_read_problem(tr_path.encode("utf-8"))
+    else:
+        tr_ptr = make_ffm_prob(tr[0], tr[1])
+
+    if va_path is not None:
+        va_ptr = ffm_read_problem(va_path.encode("utf-8"))
+    elif va is not None:
         va_ptr = make_ffm_prob(va[0], va[1])
     else:
         va_ptr = NULL

setup.py

@@ -26,7 +26,7 @@
 
 setup(
     name="ffm",
-    version="0.3.1",
+    version="0.4.0",
     description="LibFFM Python Package",
     long_description="LibFFM Python Package",
     install_requires=["numpy"],