Merge pull request bsc-wdc#51 from bsc-wdc/issue-20

Failure of 'csvm-driver' example.

examples/csvm-driver.py

@@ -7,7 +7,7 @@
 import numpy as np
 from sklearn.datasets import load_svmlight_file
 
-from dislib.data import load_file, load_files
+from dislib.data import *
 from dislib.classification import CascadeSVM
 
 from pycompss.api.api import barrier
@@ -20,30 +20,30 @@ def main():
     parser.add_argument("-dt", "--detailed_times",
                         help="get detailed execution times (read and fit)",
                         action="store_true")
-    parser.add_argument("-k", metavar="KERNEL", type=str,
+    parser.add_argument("-k", "--kernel", metavar="KERNEL", type=str,
                         help="linear or rbf (default is rbf)",
                         choices=["linear", "rbf"], default="rbf")
-    parser.add_argument("-a", metavar="CASCADE_ARITY", type=int,
+    parser.add_argument("-a", "--arity", metavar="CASCADE_ARITY", type=int,
                         help="default is 2", default=2)
-    parser.add_argument("-p", metavar="PART_SIZE", type=int,
+    parser.add_argument("-p", "--part_size", metavar="PART_SIZE", type=int,
                         help="size of the partitions in which to divide the "
                              "input dataset")
-    parser.add_argument("-i", metavar="MAX_ITERATIONS", type=int,
+    parser.add_argument("-i", "--iteration", metavar="MAX_ITERATIONS", type=int,
                         help="default is 5", default=5)
-    parser.add_argument("-g", metavar="GAMMA", type=float,
+    parser.add_argument("-g", "--gamma", metavar="GAMMA", type=float,
                         help="(only for rbf kernel) default is 1 / n_features",
                         default=None)
-    parser.add_argument("-c", metavar="C", type=float, help="default is 1",
-                        default=1)
-    parser.add_argument("-f", metavar="N_FEATURES", type=int,
+    parser.add_argument("-c", metavar="C", type=float, default=1,
+                        help="Penalty parameter C of the error term. Default:1")
+    parser.add_argument("-f", "--features", metavar="N_FEATURES", type=int,
                         help="mandatory if --libsvm option is used and "
                              "train_data is a directory (optional otherwise)",
                         default=None)
-    parser.add_argument("-t", metavar="TEST_FILE_PATH",
+    parser.add_argument("-t", "--test-file", metavar="TEST_FILE_PATH",
                         help="test CSV file path", type=str, required=False)
-    parser.add_argument("-o", metavar="OUTPUT_FILE_PATH",
+    parser.add_argument("-o", "--output_file", metavar="OUTPUT_FILE_PATH",
                         help="output file path", type=str, required=False)
-    parser.add_argument("-nd", metavar="N_DATASETS", type=int,
+    parser.add_argument("-nd", "--n_datasets", metavar="N_DATASETS", type=int,
                         help="number of times to load the dataset", default=1)
     parser.add_argument("--convergence", help="check for convergence",
                         action="store_true")
@@ -58,61 +58,54 @@ def main():
 
     train_data = args.train_data
 
-    if not args.g:
+    if not args.gamma:
         gamma = "auto"
     else:
-        gamma = args.g
-
-    if args.libsvm:
-        fmt = "libsvm"
-    else:
-        fmt = "labeled"
+        gamma = args.gamma
 
     data = []
 
     s_time = time.time()
 
     if os.path.isdir(train_data):
-        for _ in range(args.nd):
-            data.append(load_files(path=train_data, fmt=fmt,
-                                   n_features=args.f, use_array=args.dense))
+        _loader_func = load_libsvm_files if args.libsvm else load_csv_files
+        for _ in range(args.n_datasets):
+            data.append(_loader_func(train_data, args.features))
     else:
-        for _ in range(args.nd):
-            data.append(load_file(path=train_data, part_size=args.p,
-                                  fmt=fmt, n_features=args.f,
-                                  use_array=args.dense))
+        _loader_func = load_libsvm_file if args.libsvm else load_csv_file
+        for _ in range(args.n_datasets):
+            data.append(_loader_func(train_data, args.part_size, args.features))
 
-    if args.dt:
+    if args.detailed_times:
         barrier()
 
-
-    csvm = CascadeSVM(cascade_arity=args.a, max_iter=args.i, c=args.c,
-                      gamma=gamma, check_convergence=args.convergence)
+    csvm = CascadeSVM(cascade_arity=args.arity, max_iter=args.iteration,
+                      c=args.c, gamma=gamma, check_convergence=args.convergence)
 
     for d in data:
         csvm.fit(d)
 
-    out = [args.k, args.a, args.p, csvm._clf_params["gamma"], args.c,
-           csvm.iterations, csvm.converged]
+    out = [args.kernel, args.arity, args.part_size, csvm._clf_params["gamma"],
+           args.c, csvm.iterations, csvm.converged]
 
     if os.path.isdir(train_data):
         n_files = os.listdir(train_data)
         out.append(len(n_files))
 
-    if args.t:
+    if args.test_file:
         if args.libsvm:
-            testx, testy = load_svmlight_file(args.t, args.f)
+            testx, testy = load_svmlight_file(args.test_file, args.features)
 
             if args.dense:
                 testx = testx.toarray()
 
             out.append(csvm.score(testx, testy))
         else:
-            test = np.loadtxt(args.t, delimiter=",", dtype=float)
+            test = np.loadtxt(args.test_file, delimiter=",", dtype=float)
             out.append(csvm.score(test[:, :-1], test[:, -1]))
 
-    if args.o:
-        with open(args.o, "ab") as f:
+    if args.output_file:
+        with open(args.output_file, "ab") as f:
             wr = csv.writer(f)
             wr.writerow(out)
     else: