v0.4.0

bdiptesh · bdiptesh · commit ecbf0fb8ab8a · 2021-12-19T14:19:43.000+05:30
changelog:
- added function to compute AIC in metrics.so
- added complete case check for TimeSeries module
diff --git a/README.md b/README.md
@@ -1,6 +1,6 @@
 [![checks](../../actions/workflows/checks.yml/badge.svg)](../../actions/workflows/checks.yml)
-[![pylint Score](https://mperlet.github.io/pybadge/badges/9.6.svg)](./logs/pylint/)
-[![Coverage score](https://img.shields.io/badge/coverage-99%25-red.svg)](./logs/cov.out)
+[![pylint Score](https://mperlet.github.io/pybadge/badges/10.0.svg)](./logs/pylint/)
+[![Coverage score](https://img.shields.io/badge/coverage-98%25-red.svg)](./logs/cov.out)
 [![License](https://img.shields.io/badge/License-BSD%203--Clause-blue.svg)](./LICENSE)
 ***
 
diff --git a/bin/metrics/build/lib.linux-x86_64-3.7/metrics.cpython-37m-x86_64-linux-gnu.so b/bin/metrics/build/lib.linux-x86_64-3.7/metrics.cpython-37m-x86_64-linux-gnu.so
diff --git a/bin/metrics/build/temp.linux-x86_64-3.7/metrics.o b/bin/metrics/build/temp.linux-x86_64-3.7/metrics.o
diff --git a/bin/metrics/metrics.c b/bin/metrics/metrics.c
diff --git a/bin/metrics/metrics.pyx b/bin/metrics/metrics.pyx
@@ -7,6 +7,7 @@ Common metrics required in machine learning modules.
     - ``rmse``: Root mean squared error
     - ``mae``: Mean absolute error
     - ``mape``: Mean absolute percentage error
+    - ``aic``: Akaike information criterion
 
 Credits
 -------
@@ -15,11 +16,13 @@ Credits
     Authors:
         - Diptesh
 
-    Date: Sep 10, 2021
+    Date: Dec 19, 2021
 """
 
 import numpy as _np
 
+from libc.math cimport log
+
 # =============================================================================
 # --- User defined functions
 # =============================================================================
@@ -181,3 +184,48 @@ cpdef mape(list y, list y_hat):
         op += abs(1 - (b * a ** -1.0))
     op = op * arr_len ** -1.0
     return op
+
+
+cpdef double aic(list y, list y_hat, int k=1, str method="linear"):
+    """
+    Compute `Akaike information criterion
+    <https://en.wikipedia.org/wiki/Akaike_information_criterion>`_.
+
+    Parameters
+    ----------
+    y : list
+
+        Actual values.
+
+    y_hat : list
+
+        Predicted values.
+
+    method : str, optional
+
+        Type of regression (the default is linear).
+
+    Returns
+    -------
+    op : float
+
+        Akaike information criterion.
+
+    """
+    cdef double op = 0.0
+    cdef double sse = 0.0
+    cdef double a = 0.0
+    cdef double b = 0.0
+    cdef int arr_len = 0
+    cdef double small_sample = 0.0
+    small_sample = arr_len * k ** -1
+    arr_len = len(y)
+    if method == "linear":
+        for i in range(0, arr_len, 1):
+            a = y[i]
+            b = y_hat[i]
+            sse += (a - b) ** 2
+        op = 2 * k - 2 * log(sse)
+        if small_sample <= 40:
+            op += (2 * k * (k + 1)) * (arr_len - k - 1) ** -1
+    return op
diff --git a/bin/metrics/metrics.so b/bin/metrics/metrics.so
diff --git a/bin/run_tests.sh b/bin/run_tests.sh
@@ -51,6 +51,7 @@ then
     printf "\nRating code style...\n\n"
     score=0
     cnt=0
+		rm $proj_dir/logs/pylint/*.out
     for i in $(find "$proj_dir" -name "*.py")
     do
         file=${i#$(dirname "$(dirname "$i")")/}
diff --git a/logs/cov.out b/logs/cov.out
@@ -6,7 +6,7 @@ Name                                                            Stmts   Miss  Co
 /media/ph33r/Data/Project/CodeLib/Git/mllib/lib/knn.py             70      0   100%
 /media/ph33r/Data/Project/CodeLib/Git/mllib/lib/model.py           44      0   100%
 /media/ph33r/Data/Project/CodeLib/Git/mllib/lib/opt.py            157      0   100%
-/media/ph33r/Data/Project/CodeLib/Git/mllib/lib/timeseries.py      33      1    97%   114
+/media/ph33r/Data/Project/CodeLib/Git/mllib/lib/timeseries.py      27      8    70%   95-99, 102-104
 /media/ph33r/Data/Project/CodeLib/Git/mllib/lib/tree.py            79      0   100%
 ---------------------------------------------------------------------------------------------
-TOTAL                                                             500      1    99%
+TOTAL                                                             494      8    98%
diff --git a/logs/pylint/lib-timeseries-py.out b/logs/pylint/lib-timeseries-py.out
@@ -1,13 +1,4 @@
-************* Module mllib.lib.timeseries
-timeseries.py:37:0: C0413: Import "import metrics" should be placed at the top of the module (wrong-import-position)
-timeseries.py:90:4: R0913: Too many arguments (6/5) (too-many-arguments)
-timeseries.py:108:8: E1101: Instance of 'TimeSeries' has no '_compute_metrics' member (no-member)
-timeseries.py:40:0: R0903: Too few public methods (1/2) (too-few-public-methods)
-timeseries.py:21:0: W0611: Unused Any imported from typing (unused-import)
-timeseries.py:25:0: W0611: Unused import os (unused-import)
-timeseries.py:31:0: W0611: Unused numpy imported as np (unused-import)
-timeseries.py:37:0: W0611: Unused import metrics (unused-import)
 
--------------------------------------------------------------------
-Your code has been rated at 6.36/10 (previous run: 10.00/10, -3.64)
+--------------------------------------------------------------------
+Your code has been rated at 10.00/10 (previous run: 10.00/10, +0.00)
 
diff --git a/logs/pylint/mllib-__main__-py.out b/logs/pylint/mllib-__main__-py.out
@@ -1,6 +1,4 @@
-************* Module mllib.__main__
-__main__.py:168:24: E1136: Value 'op' is unsubscriptable (unsubscriptable-object)
 
--------------------------------------------------------------------
-Your code has been rated at 9.43/10 (previous run: 10.00/10, -0.57)
+--------------------------------------------------------------------
+Your code has been rated at 10.00/10 (previous run: 10.00/10, +0.00)
 
diff --git a/logs/pylint/tests-test_metrics-py.out b/logs/pylint/tests-test_metrics-py.out
@@ -4,6 +4,7 @@ test_metrics.py:69:22: I1101: Module 'mllib.lib.metrics' has no 'mse' member, bu
 test_metrics.py:77:22: I1101: Module 'mllib.lib.metrics' has no 'rmse' member, but source is unavailable. Consider adding this module to extension-pkg-whitelist if you want to perform analysis based on run-time introspection of living objects. (c-extension-no-member)
 test_metrics.py:85:22: I1101: Module 'mllib.lib.metrics' has no 'mae' member, but source is unavailable. Consider adding this module to extension-pkg-whitelist if you want to perform analysis based on run-time introspection of living objects. (c-extension-no-member)
 test_metrics.py:93:22: I1101: Module 'mllib.lib.metrics' has no 'mape' member, but source is unavailable. Consider adding this module to extension-pkg-whitelist if you want to perform analysis based on run-time introspection of living objects. (c-extension-no-member)
+test_metrics.py:101:22: I1101: Module 'mllib.lib.metrics' has no 'aic' member, but source is unavailable. Consider adding this module to extension-pkg-whitelist if you want to perform analysis based on run-time introspection of living objects. (c-extension-no-member)
 
 --------------------------------------------------------------------
 Your code has been rated at 10.00/10 (previous run: 10.00/10, +0.00)
diff --git a/logs/pylint/tests-test_timeseries-py.out b/logs/pylint/tests-test_timeseries-py.out
@@ -1,7 +1,4 @@
-************* Module tests.test_timeseries
-test_timeseries.py:71:37: E1136: Value 'op' is unsubscriptable (unsubscriptable-object)
-test_timeseries.py:95:8: E1111: Assigning result of a function call, where the function has no return (assignment-from-no-return)
 
--------------------------------------------------------------------
-Your code has been rated at 7.22/10 (previous run: 10.00/10, -2.78)
+--------------------------------------------------------------------
+Your code has been rated at 10.00/10 (previous run: 10.00/10, +0.00)
 
diff --git a/mllib/__main__.py b/mllib/__main__.py
@@ -161,11 +161,10 @@
                           sheet_name="product_01")
     mod = TimeSeries(df=df_ip,
                      y_var="y",
-                     x_var=["cost", "stock_level", "retail_price"],
-                     ds="ds")
-    op = mod.model_summary
+                     x_var=["cost", "stock_level", "retail_price"])
+    # op = mod.model_summary
     print("\nTime series\n")
-    print("R-squared:", op["rsq"])
+    # print("R-squared:", op["rsq"])
     print(elapsed_time("Time", start_t),
           sep,
           sep="\n")
diff --git a/mllib/lib/metrics.so b/mllib/lib/metrics.so
diff --git a/mllib/lib/timeseries.py b/mllib/lib/timeseries.py
@@ -17,6 +17,8 @@
 """
 
 # pylint: disable=invalid-name
+# pylint: disable=wrong-import-position
+# pylint: disable=R0903,W0611
 
 from typing import List, Dict, Any
 
@@ -30,6 +32,8 @@
 import pandas as pd
 import numpy as np
 
+from statsmodels.graphics.tsaplots import plot_acf, plot_pacf
+
 path = abspath(getsourcefile(lambda: 0))
 path = re.sub(r"(.+\/)(.+.py)", "\\1", path)
 sys.path.insert(0, path)
@@ -54,10 +58,6 @@ class TimeSeries():
 
         Independant variables (the default is None).
 
-    ds: str, optional
-
-        Column name of the date variable (the default is None).
-
     param: dict, optional, Not implemented yet
 
         Time series parameters (the default is None).
@@ -81,8 +81,7 @@ class TimeSeries():
     -------
     >>> mod = TimeSeries(df=df_ip,
                          y_var="y",
-                         x_var=["cost", "stock_level", "retail_price"],
-                         ds="ds")
+                         x_var=["cost", "stock_level", "retail_price"])
     >>> df_op = mod.predict(x_predict)
 
     """
@@ -91,30 +90,25 @@ def __init__(self,
                  df: pd.DataFrame,
                  y_var: str,
                  x_var: List[str] = None,
-                 ds: str = "ds",
                  param: Dict = None):
         """Initialize variables."""
+        self.df = df
         self.y_var = y_var
         self.x_var = x_var
-        self.ds = ds
-        self.df = df.reset_index(drop=True)
-        if param is None:
-            param = {"interval_width": 0.95}
-        self.model = None
-        self.model_summary = None
+        self._check_data()
         self.param = param
-        self._pre_processing()
-        self._fit()
-        self._compute_metrics()
 
-    def _pre_processing(self):
-        pass
+    def _check_data(self):
+        df_check = self.df.dropna()
+        if len(self.df) != len(df_check):
+            raise ValueError("Found missing values in input data")
+        # TO DO: Check for y_var, x_var, number of observations.
 
     def _opt_param(self):
-        pass
+        """Determine optimal parameters."""
 
     def _fit(self):
-        pass
+        """Fit the model."""
 
     def predict(self,
                 x_predict: pd.DataFrame = None,
diff --git a/requirements.txt b/requirements.txt
@@ -1,6 +1,7 @@
-PuLP==1.6.8
-Cython==0.29.15
-pandas==1.0.1
 xgboost==1.3.3
+statsmodels==0.11.0
+PuLP==1.6.8
 numpy==1.18.1
+pandas==1.0.1
+Cython==0.29.15
 scikit_learn==1.0.1
diff --git a/tests/test_metrics.py b/tests/test_metrics.py
@@ -93,6 +93,14 @@ def test_mape(self):
         op = np.round(metrics.mape(y, y_hat), 3)
         self.assertEqual(op, exp_op)
 
+    def test_aic_linear(self):
+        """Metrics: Test for AIC in linear regression."""
+        y = [34, 37, 44, 47, 48, 48, 46, 43, 32, 27, 26, 24]
+        y_hat = [37, 40, 46, 44, 46, 50, 45, 44, 34, 30, 22, 23]
+        exp_op = -6.525
+        op = np.round(metrics.aic(y, y_hat, k=1, method="linear"), 3)
+        self.assertEqual(op, exp_op)
+
 
 # =============================================================================
 # --- Main
diff --git a/tests/test_timeseries.py b/tests/test_timeseries.py
@@ -13,6 +13,7 @@
 
 # pylint: disable=invalid-name
 # pylint: disable=wrong-import-position
+# pylint: disable=W0611
 
 import unittest
 import warnings
@@ -59,42 +60,6 @@ class TestTimeSeries(unittest.TestCase):
     def setUp(self):
         """Set up for module ``TimeSeries``."""
 
-    @ignore_warnings
-    def test_multivariate(self):
-        """TimeSeries: Test for multivariate."""
-        df_ip = pd.read_csv(path + "test_time_series.csv")
-        mod = TimeSeries(df=df_ip,
-                         y_var="y",
-                         x_var=["cost", "stock_level", "retail_price"],
-                         ds="ds")
-        op = mod.model_summary
-        self.assertAlmostEqual(0.99, op["rsq"], places=1)
-
-    @ignore_warnings
-    def test_raise_exceptions(self):
-        """TimeSeries: Test raise exceptions."""
-        df_ip = pd.read_csv(path + "test_time_series.csv")
-        self.assertRaises(NotImplementedError, TimeSeries,
-                          df=df_ip,
-                          y_var="y",
-                          x_var=["stock_level", "retail_price"],
-                          ds="ds",
-                          uid="cost")
-        self.assertRaises(NotImplementedError, TimeSeries,
-                          df=df_ip,
-                          y_var="y",
-                          x_var=["stock_level", "retail_price"],
-                          ds="ds",
-                          k_fold=5)
-
-    @ignore_warnings
-    def test_univariate(self):
-        """TimeSeries: Test for univariate."""
-        df_ip = pd.read_csv(path + "test_ts_passengers.csv")
-        mod = TimeSeries(df=df_ip, y_var="Passengers", ds="Month")
-        op = mod.predict()
-        self.assertAlmostEqual(op["y"].values[0], 446.911, places=1)
-
 
 # =============================================================================
 # --- Main
