From df582d235a6e6c8e114053015a7b7392bee8f570 Mon Sep 17 00:00:00 2001
From: Chris Holden <ceholden@gmail.com>
Date: Sun, 6 Dec 2015 17:03:02 -0500
Subject: [PATCH] Add regression diagnotic calcs #70

---
 tests/regression/conftest.py                  |  6 +++++
 .../regression/test_regression_diagnostics.py | 18 +++++++++++++++
 yatsm/regression/diagnostics.py               | 22 +++++++++++++++++++
 3 files changed, 46 insertions(+)
 create mode 100644 tests/regression/test_regression_diagnostics.py
 create mode 100644 yatsm/regression/diagnostics.py

diff --git a/tests/regression/conftest.py b/tests/regression/conftest.py
index b33362ab..e4de9069 100644
--- a/tests/regression/conftest.py
+++ b/tests/regression/conftest.py
@@ -1,11 +1,17 @@
 import os
 
+import numpy as np
 import pandas as pd
 import pytest
 
 here = os.path.dirname(__file__)
 
 
+@pytest.fixture(scope='function')
+def prng():
+    return np.random.RandomState(123456789)
+
+
 @pytest.fixture(scope='function')
 def airquality(request):
     airquality = pd.read_csv(os.path.join(here, 'data', 'airquality.csv'))
diff --git a/tests/regression/test_regression_diagnostics.py b/tests/regression/test_regression_diagnostics.py
new file mode 100644
index 00000000..efe5c3f2
--- /dev/null
+++ b/tests/regression/test_regression_diagnostics.py
@@ -0,0 +1,18 @@
+import numpy as np
+import pytest
+
+from yatsm.regression.diagnostics import rmse
+
+n = 500
+
+
+@pytest.fixture
+def y(prng):
+    y = prng.standard_normal(n)
+    return y
+
+
+def test_rmse(y):
+    yhat = y + 1.0
+    _rmse = rmse(y, yhat)
+    np.testing.assert_allclose(_rmse, 1.0)
diff --git a/yatsm/regression/diagnostics.py b/yatsm/regression/diagnostics.py
new file mode 100644
index 00000000..85499ce3
--- /dev/null
+++ b/yatsm/regression/diagnostics.py
@@ -0,0 +1,22 @@
+""" Regression diagnostics calculations
+
+Includes:
+    - rmse: calculate root mean squared error
+"""
+import numpy as np
+
+from ..accel import try_jit
+
+
+@try_jit(nopython=True)
+def rmse(y, yhat):
+    """ Calculate and return Root Mean Squared Error (RMSE)
+
+    Args:
+        y (np.ndarray): known values
+        yhat (np.ndarray): predicted values
+
+    Returns:
+        float: Root Mean Squared Error
+    """
+    return ((y - yhat) ** 2).mean() ** 0.5