Collect Ideas for the PyAF 5.0 roadmap #203. Some tests for Business …

…Days/Hours

tests/business_days_hours/test_ozone_on_business_days.py

@@ -0,0 +1,60 @@
+import pandas as pd
+import numpy as np
+import pyaf.ForecastEngine as autof
+import pyaf.Bench.TS_datasets as tsds
+
+#get_ipython().magic('matplotlib inline')
+
+b1 = tsds.load_ozone()
+df = b1.mPastData
+
+
+df[b1.mTimeVar + '_BusinessDaily'] = pd.bdate_range('2000-1-1', periods=df.shape[0])
+
+
+#df.to_csv("outputs/ozone_WDHMS.csv");
+#df.tail(10)
+#df[:-10].tail()
+#df[:-10:-1]
+#df.describe()
+
+
+timevar = b1.mTimeVar +  '_BusinessDaily'
+
+lEngine = autof.cForecastEngine()
+lEngine
+
+H = b1.mHorizon;
+# lEngine.mOptions.enable_slow_mode();
+# lEngine.mOptions.mDebugPerformance = True;
+lEngine.mOptions.set_active_autoregressions([]);
+
+lEngine.train(df , timevar , b1.mSignalVar, H);
+lEngine.getModelInfo();
+print(lEngine.mSignalDecomposition.mTrPerfDetails.head());
+
+lEngine.mSignalDecomposition.mBestModel.mTimeInfo.mResolution
+
+lEngine.standardPlots("outputs/my_ozone_bday");
+
+
+dfapp_in = df.copy();
+dfapp_in.tail()
+
+# H = 12
+dfapp_out = lEngine.forecast(dfapp_in, H);
+#dfapp_out.to_csv("outputs/ozone_" + timevar + "apply_out.csv")
+dfapp_out.tail(2 * H)
+print("Forecast Columns " , dfapp_out.columns);
+Forecast_DF = dfapp_out[[timevar , b1.mSignalVar, b1.mSignalVar + '_Forecast']]
+print(Forecast_DF.info())
+print("Forecasts\n" , Forecast_DF.tail(2*H));
+
+print("\n\n<ModelInfo>")
+print(lEngine.to_json());
+print("</ModelInfo>\n\n")
+print("\n\n<Forecast>")
+print(Forecast_DF.tail(2*H).to_json(date_format='iso'))
+print("</Forecast>\n\n")
+
+# lEngine.standardPlots(name = "outputs/ozone_" + timevar)

tests/business_days_hours/test_ozone_on_business_hours.py

@@ -0,0 +1,59 @@
+import pandas as pd
+import numpy as np
+import pyaf.ForecastEngine as autof
+import pyaf.Bench.TS_datasets as tsds
+
+#get_ipython().magic('matplotlib inline')
+
+b1 = tsds.load_ozone()
+df = b1.mPastData
+
+
+df[b1.mTimeVar + '_BusinessHourly'] = pd.bdate_range('2000-1-1', periods=df.shape[0], freq = '2BH')
+
+#df.to_csv("outputs/ozone_WDHMS.csv");
+#df.tail(10)
+#df[:-10].tail()
+#df[:-10:-1]
+#df.describe()
+
+
+timevar = b1.mTimeVar +  '_BusinessHourly'
+
+lEngine = autof.cForecastEngine()
+lEngine
+
+H = b1.mHorizon;
+# lEngine.mOptions.enable_slow_mode();
+# lEngine.mOptions.mDebugPerformance = True;
+lEngine.mOptions.set_active_autoregressions([]);
+
+lEngine.train(df , timevar , b1.mSignalVar, H);
+lEngine.getModelInfo();
+print(lEngine.mSignalDecomposition.mTrPerfDetails.head());
+
+lEngine.mSignalDecomposition.mBestModel.mTimeInfo.mResolution
+
+lEngine.standardPlots("outputs/my_ozone_bhour");
+
+
+dfapp_in = df.copy();
+dfapp_in.tail()
+
+# H = 12
+dfapp_out = lEngine.forecast(dfapp_in, H);
+#dfapp_out.to_csv("outputs/ozone_" + timevar + "apply_out.csv")
+dfapp_out.tail(2 * H)
+print("Forecast Columns " , dfapp_out.columns);
+Forecast_DF = dfapp_out[[timevar , b1.mSignalVar, b1.mSignalVar + '_Forecast']]
+print(Forecast_DF.info())
+print("Forecasts\n" , Forecast_DF.tail(2*H));
+
+print("\n\n<ModelInfo>")
+print(lEngine.to_json());
+print("</ModelInfo>\n\n")
+print("\n\n<Forecast>")
+print(Forecast_DF.tail(2*H).to_json(date_format='iso'))
+print("</Forecast>\n\n")
+
+# lEngine.standardPlots(name = "outputs/ozone_" + timevar)