taxi

commit 8b27690c8d77585f173412e5719787c48272674e
parent fe704521e6cb4d7f32414b55044b6e2240524bf5
Author: Alex Auvolat <alex.auvolat@ens.fr>
Date:   Wed, 29 Apr 2015 19:16:22 -0400

Fix CSV id normalization

Diffstat:
M
data.py
 | 
29
+++++++++++++++++++++--------
M
model.py
 | 
6
+++---

2 files changed, 24 insertions(+), 11 deletions(-)
diff --git a/data.py b/data.py
@@ -13,22 +13,35 @@ if socket.gethostname() == "adeb.laptop":
 else:
     DATA_PATH="/data/lisatmp3/auvolat/taxikaggle"
 
-client_ids = {int(x): y+1 for y, x in enumerate(open(DATA_PATH+"/client_ids.txt"))}
-
-def get_client_id(n):
-    if n in client_ids:
-        return client_ids[n]
-    else:
-        return 0
+H5DATA_PATH = '/data/lisatmp3/simonet/taxi/data.hdf5'
 
 porto_center = numpy.array([41.1573, -8.61612], dtype=theano.config.floatX)
 data_std = numpy.sqrt(numpy.array([0.00549598, 0.00333233], dtype=theano.config.floatX))
 
-n_clients = 57124 #57105
+n_clients = 57124
+n_train_clients = 57105
 n_stands = 63
 
 dataset_size = 1710670
 
+# ---- Read client IDs and create reverse dictionnary
+
+def make_client_ids():
+    f = h5py.File(H5DATA_PATH, "r")
+    l = f['uniq_origin_call']
+    r = {}
+    for i in range(l.shape[0]):
+        r[l[i]] = i
+    return r
+
+client_ids = make_client_ids()
+
+def get_client_id(n):
+    if n in client_ids:
+        return client_ids[n]
+    else:
+        return 0
+
 class CallType(Enum):
     CENTRAL = 0
     STAND = 1
diff --git a/model.py b/model.py
@@ -46,7 +46,7 @@ if __name__ == "__main__":
 
 def setup_stream():
     # Load the training and test data
-    train = H5PYDataset('/data/lisatmp3/simonet/taxi/data.hdf5',
+    train = H5PYDataset(H5DATA_PATH,
                         which_set='train',
                         subset=slice(0, data.dataset_size - config.n_valid),
                         load_in_memory=True)
@@ -59,7 +59,7 @@ def setup_stream():
                                         'destination_latitude', 'destination_longitude'))
     train_stream = Batch(train, iteration_scheme=ConstantScheme(config.batch_size))
 
-    valid = H5PYDataset('/data/lisatmp3/simonet/taxi/data.hdf5',
+    valid = H5PYDataset(H5DATA_PATH,
                         which_set='train',
                         subset=slice(data.dataset_size - config.n_valid, data.dataset_size),
                         load_in_memory=True)
@@ -106,7 +106,7 @@ def main():
     # x_lastk_longitude = theano.printing.Print("x_lastk_longitude")(x_lastk_longitude)
 
     # Define the model
-    client_embed_table = LookupTable(length=data.n_clients+1, dim=config.dim_embed, name='client_lookup')
+    client_embed_table = LookupTable(length=data.n_train_clients+1, dim=config.dim_embed, name='client_lookup')
     stand_embed_table = LookupTable(length=data.n_stands+1, dim=config.dim_embed, name='stand_lookup')
     mlp = MLP(activations=[Rectifier() for _ in config.dim_hidden] + [Identity()],
                        dims=[config.dim_input] + config.dim_hidden + [config.dim_output])