Update and release a new version

Author: moisesrsantos

.github/workflows/pypi-release.yml

@@ -0,0 +1,35 @@
+name: Upload to PyPI
+
+on:
+  release:
+    types: [published]
+
+permissions:
+  contents: read
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.9'
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install build twine
+
+      - name: Build package
+        run: python -m build
+
+      - name: Publish to PyPI
+        env:
+          TWINE_USERNAME: __token__
+          TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
+        run: twine upload dist/*

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "tsmorph"
-version = "0.1.0"
+version = "0.1.1"
 description = "A package for generating semi-synthetic time series using morphing techniques."
 readme = "README.md"
 authors = [

setup.py

@@ -2,7 +2,7 @@
 
 setup(
     name='tsmorph',
-    version='0.1.0',
+    version='0.1.1',
     packages=find_packages(),
     install_requires=[
         'numpy',

tsmorph/tsmorph.py

@@ -2,8 +2,7 @@
 import pandas as pd
 import matplotlib.pyplot as plt
 from pycatch22 import catch22_all
-from neuralforecast.losses.numpy import mase
-from .utils import plot_gradient_timeseries
+from .utils import plot_gradient_timeseries, nmae
 
 class TSmorph:
     """
@@ -61,7 +60,7 @@ def analyze_morph_performance(self, df: pd.DataFrame, model, horizon: int, seaso
             horizon (int): Forecast horizon for testing.
         """
         feature_values = []
-        mase_values = []
+        nmae_values = []
 
         for col in df.columns:
             series = df[col].values
@@ -80,19 +79,19 @@ def analyze_morph_performance(self, df: pd.DataFrame, model, horizon: int, seaso
             forecast_df = model.predict(test)
             forecast = forecast_df[forecast_df['unique_id'] == col][model.models[0].__class__.__name__].values[:horizon]
 
-            mase_values.append(mase(y=test['y'].values, y_hat=forecast, y_train=df_forecast.iloc[:-horizon]['y'].values, seasonality=seasonality))
+            nmae_values.append(nmae(y=test['y'].values, y_hat=forecast))
 
         feature_values = np.array(feature_values)
-        mase_values = np.array(mase_values)
+        nmae_values = np.array(nmae_values)
 
         num_features = feature_values.shape[1]
         x_values = np.arange(len(df.columns))
 
         for i in range(num_features):
             plt.figure(figsize=(8, 5))
-            sc = plt.scatter(x_values, feature_values[:, i], c=mase_values, cmap='viridis', edgecolors='k')
-            plt.colorbar(sc, label='MASE')
+            sc = plt.scatter(x_values, feature_values[:, i], c=nmae_values, cmap='viridis', edgecolors='k')
+            plt.colorbar(sc, label='NMAE')
             plt.xlabel('Granularity Level')
             plt.ylabel(feature_names[i])
-            plt.title(f'{feature_names[i]} variation with MASE')
+            plt.title(f'{feature_names[i]} variation with NMAE')
             plt.show()

tsmorph/utils.py

@@ -1,4 +1,5 @@
 import matplotlib.pyplot as plt
+import numpy as np
 
 def plot_gradient_timeseries(df, start_color='#61E6AA', end_color='#5722B1'):
     # Convert hex to RGB
@@ -36,4 +37,17 @@ def hex_to_rgb(hex_color):
     # Adjust layout to prevent label cutoff
     plt.tight_layout()
     plt.show()
-    return
+    return
+
+def nmae(y, y_hat):
+    """
+    Computes the Normalized Mean Absolute Error (NMAE).
+
+    Args:
+        y (np.array): True values.
+        y_hat (np.array): Predicted values.
+    
+    Returns:
+        float: NMAE score.
+    """
+    return np.mean(np.abs(y - y_hat)) / np.mean(np.abs(y))