-
Notifications
You must be signed in to change notification settings - Fork 73
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
5 changed files
with
172 additions
and
26 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,103 @@ | ||
| from __future__ import absolute_import | ||
|
|
||
| import pandas as pd | ||
| import numpy as np | ||
|
|
||
| import pyaf.ForecastEngine as autof | ||
|
|
||
|
|
||
| def draw(iEngine, iFormula, iSignal): | ||
| print("DRAWING_PREDICTION_INTERVALS", iFormula) | ||
| b64_string = iEngine.mSignalDecomposition.mBestModels[iSignal].getPredictionIntervalPlot() | ||
| import io, base64, imageio | ||
| sourceString = io.BytesIO(base64.b64decode(b64_string)) | ||
| lArray = imageio.imread(sourceString, pilmode='RGB') | ||
| return lArray | ||
|
|
||
| def plot_model(arg): | ||
| (lDataset , lSpec) = arg | ||
| lSignal = lDataset.mSignalVar | ||
| print("PLOT_MODEL_START" , lSpec) | ||
| lEngine = autof.cForecastEngine() | ||
| lEngine | ||
| H = lDataset.mHorizon; | ||
| # lEngine.mOptions.enable_slow_mode(); | ||
| # lEngine.mOptions.mDebugPerformance = True; | ||
| lEngine.mOptions.mNbCores = 1 | ||
| lEngine.mOptions.set_active_transformations(lSpec["Transformation"]); | ||
| lEngine.mOptions.set_active_trends(lSpec["Trend"]); | ||
| lEngine.mOptions.set_active_periodics(lSpec["Periodics"]); | ||
| lEngine.mOptions.set_active_autoregressions(lSpec["AutoRegression"]); | ||
| lEngine.mOptions.set_active_decomposition_types(lSpec["Decomposition"]); | ||
|
|
||
| lPlots = {} | ||
| lDict = {} | ||
| try: | ||
| lEngine.train(lDataset.mPastData , lDataset.mTimeVar , lDataset.mSignalVar, H); | ||
| lEngine.getModelInfo(); | ||
| lDict = lEngine.to_dict() | ||
| lMAPE = lDict[lSignal]["Model_Performance"]["MAPE"] | ||
| lFormula = lDict[lSignal]["Model"]["Best_Decomposition"] | ||
| print("PLOT_MODEL_END" , lSpec , | ||
| lFormula, lMAPE) | ||
| img = draw(lEngine, lFormula, lSignal) | ||
| except Exception as e: | ||
| print("PLOT_MODEL_END_FAILED" , lSpec, str(e)) | ||
| return (lSpec, None, None, None) | ||
|
|
||
| return (lSpec, img, lFormula, lMAPE) | ||
|
|
||
|
|
||
| class cModelEstheticsVisualizer: | ||
|
|
||
| def __init__(self): | ||
| self.mDataset = None | ||
| self.mNbThreads = 12 | ||
| pass | ||
|
|
||
|
|
||
| def gen_all(self): | ||
| lEngine = autof.cForecastEngine() | ||
| lSpecs = [] | ||
| lSpec = {} | ||
| lKnownAutoRegressions = [x for x in lEngine.mOptions.mKnownAutoRegressions if not x.endswith('X')] | ||
| lKnownAutoRegressions = [x for x in lKnownAutoRegressions if (x != 'CROSTON')] | ||
| lKnownPeriodics = ['NoCycle', 'BestCycle', 'Seasonal_MonthOfYear']; | ||
| for tr in lEngine.mOptions.mKnownTransformations: | ||
| lSpec["Transformation"] = tr | ||
| for tr1 in lEngine.mOptions.mKnownTrends: | ||
| lSpec["Trend"] = tr1 | ||
| for per in lKnownPeriodics: | ||
| lSpec["Periodics"] = per | ||
| for ar in lKnownAutoRegressions: | ||
| lSpec["AutoRegression"] = ar | ||
| for dec in lEngine.mOptions.mKnownDecompositionTypes: | ||
| lSpec["Decomposition"] = dec | ||
| lSpecs = lSpecs + [(self.mDataset, lSpec.copy())] | ||
| print("TESTED_MODELS" , len(lSpecs)) | ||
| lPlots = {} | ||
| from multiprocessing import Pool | ||
| pool = Pool(self.mNbThreads) | ||
| for res in pool.imap(plot_model, lSpecs): | ||
| (lSpec, img, lFormula, lMAPE) = res | ||
| lPlots[str(lSpec)] = (lSpec, img, lFormula, lMAPE) | ||
| pool.close() | ||
| pool.join() | ||
| return lPlots | ||
|
|
||
| def generate_video(self, iDataset): | ||
| self.mDataset = iDataset | ||
| lSignal = self.mDataset.mSignalVar | ||
| plots = self.gen_all() | ||
| images = [] | ||
| for(lSpec , lValue) in plots.items(): | ||
| (lSpec1, img, lFormula, lMAPE) = lValue | ||
| images.append((img , lMAPE)) | ||
|
|
||
| images = sorted(images, key = lambda x : -x[1]) | ||
|
|
||
| import imageio as iio | ||
| writer = iio.get_writer(lSignal + "_models.mp4", fps=20) | ||
| for im in images: | ||
| writer.append_data(im[0]) | ||
| writer.close() |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,10 @@ | ||
|
|
||
|
|
||
| import pyaf.Bench.TS_datasets as tsds | ||
| import model_esthetics_visualizer as viz | ||
|
|
||
| lDataset = tsds.load_airline_passengers() | ||
|
|
||
| lVisualizer = viz.cModelEstheticsVisualizer() | ||
|
|
||
| lVisualizer.generate_video(lDataset) |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,10 @@ | ||
|
|
||
|
|
||
| import pyaf.Bench.TS_datasets as tsds | ||
| import model_esthetics_visualizer as viz | ||
|
|
||
| lDataset = tsds.load_ozone() | ||
|
|
||
| lVisualizer = viz.cModelEstheticsVisualizer() | ||
|
|
||
| lVisualizer.generate_video(lDataset) |