fixed rank build bug

alphamind/benchmarks/data/neutralize.py

@@ -18,7 +18,7 @@ def benchmark_neutralize(n_samples: int, n_features: int, n_loops: int) -> None:
     print("Starting least square fitting benchmarking")
     print("Parameters(n_samples: {0}, n_features: {1}, n_loops: {2})".format(n_samples, n_features, n_loops))
 
-    y = np.random.randn(n_samples)
+    y = np.random.randn(n_samples, 5)
     x = np.random.randn(n_samples, n_features)
 
     start = dt.datetime.now()
@@ -32,7 +32,7 @@ def benchmark_neutralize(n_samples: int, n_features: int, n_loops: int) -> None:
     for _ in range(n_loops):
         benchmark_model = LinearRegression(fit_intercept=False)
         benchmark_model.fit(x, y)
-        _ = y - x @ benchmark_model.coef_
+        _ = y - x @ benchmark_model.coef_.T
     benchmark_model_time = dt.datetime.now() - start
 
     print('{0:20s}: {1}'.format('Benchmark model', benchmark_model_time))

alphamind/benchmarks/portfolio/rankbuild.py

@@ -16,7 +16,7 @@ def benchmark_build_rank(n_samples: int, n_loops: int, n_included: int) -> None:
     print("Starting portfolio construction by rank benchmarking")
     print("Parameters(n_samples: {0}, n_included: {1}, n_loops: {2})".format(n_samples, n_included, n_loops))
 
-    x = np.random.randn(n_samples)
+    x = np.random.randn(n_samples, 1)
 
     start = dt.datetime.now()
     for _ in range(n_loops):
@@ -27,8 +27,8 @@ def benchmark_build_rank(n_samples: int, n_loops: int, n_included: int) -> None:
 
     start = dt.datetime.now()
     for _ in range(n_loops):
-        expected_weights = np.zeros(len(x))
-        expected_weights[(-x).argsort().argsort() < n_included] = 1. / n_included
+        expected_weights = np.zeros((len(x), 1))
+        expected_weights[(-x).argsort(axis=0).argsort(axis=0) < n_included] = 1. / n_included
     benchmark_model_time = dt.datetime.now() - start
 
     print('{0:20s}: {1}'.format('Benchmark model', benchmark_model_time))
@@ -39,7 +39,7 @@ def benchmark_build_rank_with_group(n_samples: int, n_loops: int, n_included: in
     print("Starting  portfolio construction by rank with group-by values benchmarking")
     print("Parameters(n_samples: {0}, n_included: {1}, n_loops: {2}, n_groups: {3})".format(n_samples, n_included, n_loops, n_groups))
 
-    x = np.random.randn(n_samples)
+    x = np.random.randn(n_samples, 1)
     groups = np.random.randint(n_groups, size=n_samples)
 
     start = dt.datetime.now()
@@ -51,8 +51,8 @@ def benchmark_build_rank_with_group(n_samples: int, n_loops: int, n_included: in
 
     start = dt.datetime.now()
     for _ in range(n_loops):
-        grouped_ordering = pd.Series(-x).groupby(groups).rank()
-        expected_weights = np.zeros(len(x))
+        grouped_ordering = pd.DataFrame(-x).groupby(groups).rank()
+        expected_weights = np.zeros((len(x), 1))
         masks = grouped_ordering <= n_included
         expected_weights[masks] = 1. / np.sum(masks)
     benchmark_model_time = dt.datetime.now() - start

alphamind/portfolio/rankbuilder.py

@@ -7,18 +7,21 @@ Created on 2017-4-26
 
 import numpy as np
 from numpy import zeros
+from numpy import arange
 
 
 def rank_build(er: np.ndarray, use_rank: int, groups: np.ndarray=None) -> np.ndarray:
-    length = len(er)
-    neg_er = -er
-    masks = zeros(length, dtype=bool)
-    weights = zeros(length)
 
+    if er.ndim == 1 or (er.shape[0] == 1 or er.shape[1] == 1):
+        """ fast path """
+        neg_er = -er.flatten()
+        length = len(neg_er)
+        weights = zeros((length, 1))
         if groups is not None:
             max_g = groups.max()
-        index_range = np.arange(length)
-        for i in range(max_g+1):
+            index_range = arange(length)
+            masks = zeros(length, dtype=bool)
+            for i in range(max_g + 1):
                 current_mask = groups == i
                 current_index = index_range[current_mask]
                 current_ordering = neg_er[current_mask].argsort()
@@ -26,8 +29,32 @@ def rank_build(er: np.ndarray, use_rank: int, groups: np.ndarray=None) -> np.nda
             weights[masks] = 1. / masks.sum()
         else:
             ordering = neg_er.argsort()
-        masks[ordering[:use_rank]] = True
-        weights[masks] = 1. / use_rank
+            weights[ordering[:use_rank]] = 1. / use_rank
+        return weights
+    else:
+        length = er.shape[0]
+        width = er.shape[1]
+        neg_er = -er
+        weights = zeros((length, width))
+
+        if groups is not None:
+            max_g = groups.max()
+            index_range = arange(length)
+            masks = zeros((length, width), dtype=bool)
+            for i in range(max_g+1):
+                current_mask = groups == i
+                current_index = index_range[current_mask]
+                current_ordering = neg_er[current_mask].argsort(axis=0)
+                for j in range(width):
+                    masks[current_index[current_ordering[:use_rank, j]], j] = True
+            choosed = masks.sum(axis=0)
+
+            for j in range(width):
+                weights[masks[:, j], j] = 1. / choosed[j]
+        else:
+            ordering = neg_er.argsort(axis=0)
+            for j in range(width):
+                weights[ordering[:use_rank, j], j] = 1. / use_rank
         return weights
 
 
@@ -35,12 +62,12 @@ if __name__ == '__main__':
 
     import datetime as dt
 
-    x = np.random.randn(4)
+    x = np.random.randn(4, 2)
 
     groups = np.random.randint(2, size=4)
 
     start = dt.datetime.now()
-    for i in range(10000):
-        weights = rank_build(x, 1, groups)
+    for i in range(100):
+        weights = rank_build(x, 1)#, groups)
     print(dt.datetime.now() - start)
 

alphamind/tests/data/test_neutralize.py

@@ -15,7 +15,7 @@ class TestNeutralize(unittest.TestCase):
 
     def test_neutralize(self):
 
-        y = np.random.randn(3000)
+        y = np.random.randn(3000, 4)
         x = np.random.randn(3000, 10)
 
         calc_res = neutralize(x, y)
@@ -23,7 +23,7 @@ class TestNeutralize(unittest.TestCase):
         model = LinearRegression(fit_intercept=False)
         model.fit(x, y)
 
-        exp_res = y - x @ model.coef_
+        exp_res = y - x @ model.coef_.T
 
         np.testing.assert_array_almost_equal(calc_res, exp_res)
 

alphamind/tests/portfolio/test_rankbuild.py

@@ -18,12 +18,15 @@ class TestRankBuild(unittest.TestCase):
         n_samples = 3000
         n_included = 300
 
-        x = np.random.randn(n_samples)
+        x = np.random.randn(n_samples, 2)
 
         calc_weights = rank_build(x, n_included)
 
-        expected_weights = np.zeros(len(x))
-        expected_weights[(-x).argsort().argsort() < n_included] = 1. / n_included
+        expected_weights = np.zeros((len(x), 2))
+        masks = (-x).argsort(axis=0).argsort(axis=0) < n_included
+
+        for j in range(x.shape[1]):
+            expected_weights[masks[:, j], j] = 1. / n_included
 
         np.testing.assert_array_almost_equal(calc_weights, expected_weights)
 
@@ -33,15 +36,17 @@ class TestRankBuild(unittest.TestCase):
         n_include = 10
         n_groups = 30
 
-        x = np.random.randn(n_samples)
+        x = np.random.randn(n_samples, 2)
         groups = np.random.randint(n_groups, size=n_samples)
 
         calc_weights = rank_build(x, n_include, groups)
 
-        grouped_ordering = pd.Series(-x).groupby(groups).rank()
-        expected_weights = np.zeros(len(x))
-        masks = grouped_ordering <= n_include
-        expected_weights[masks] = 1. / np.sum(masks)
+        grouped_ordering = pd.DataFrame(-x).groupby(groups).rank()
+        expected_weights = np.zeros((len(x), 2))
+        masks = (grouped_ordering <= n_include).values
+        choosed = masks.sum(axis=0)
+        for j in range(x.shape[1]):
+            expected_weights[masks[:, j], j] = 1. / choosed[j]
 
         np.testing.assert_array_almost_equal(calc_weights, expected_weights)