autots 0.2.2__tar.gz → 0.2.3__tar.gz

{AutoTS-0.2.2 → AutoTS-0.2.3/AutoTS.egg-info}/PKG-INFO

@@ -1,12 +1,14 @@
 Metadata-Version: 2.1
 Name: AutoTS
-Version: 0.2.2
+Version: 0.2.3
 Summary: Automated Time Series Forecasting
 Home-page: https://github.com/winedarksea/AutoTS
 Author: Colin Catlin
 Author-email: colin.catlin@syllepsis.live
 License: MIT
 Description: # AutoTS
+        [![PyPI download month](https://img.shields.io/pypi/dm/autots.svg)](https://pypi.python.org/pypi/autots/)
+        [![PyPI version fury.io](https://badge.fury.io/py/autots.svg)](https://pypi.python.org/pypi/autots/)
         
         <img src="/img/autots_logo.png" width="400" height="184" title="AutoTS Logo">
         

{AutoTS-0.2.2 → AutoTS-0.2.3}/AutoTS.egg-info/SOURCES.txt

@@ -38,6 +38,7 @@ autots/templates/__init__.py
 autots/templates/general.py
 autots/templates/generic_template.csv
 autots/tools/__init__.py
+autots/tools/cpu_count.py
 autots/tools/hierarchial.py
 autots/tools/holiday.py
 autots/tools/impute.py

{AutoTS-0.2.2 → AutoTS-0.2.3}/AutoTS.egg-info/requires.txt

@@ -12,3 +12,4 @@ gluonts
 tensorflow
 xgboost
 lightgbm
+psutil

{AutoTS-0.2.2/AutoTS.egg-info → AutoTS-0.2.3}/PKG-INFO

@@ -1,12 +1,14 @@
 Metadata-Version: 2.1
 Name: AutoTS
-Version: 0.2.2
+Version: 0.2.3
 Summary: Automated Time Series Forecasting
 Home-page: https://github.com/winedarksea/AutoTS
 Author: Colin Catlin
 Author-email: colin.catlin@syllepsis.live
 License: MIT
 Description: # AutoTS
+        [![PyPI download month](https://img.shields.io/pypi/dm/autots.svg)](https://pypi.python.org/pypi/autots/)
+        [![PyPI version fury.io](https://badge.fury.io/py/autots.svg)](https://pypi.python.org/pypi/autots/)
         
         <img src="/img/autots_logo.png" width="400" height="184" title="AutoTS Logo">
         

{AutoTS-0.2.2 → AutoTS-0.2.3}/README.md

@@ -1,4 +1,6 @@
 # AutoTS
+[![PyPI download month](https://img.shields.io/pypi/dm/autots.svg)](https://pypi.python.org/pypi/autots/)
+[![PyPI version fury.io](https://badge.fury.io/py/autots.svg)](https://pypi.python.org/pypi/autots/)
 
 <img src="/img/autots_logo.png" width="400" height="184" title="AutoTS Logo">
 

{AutoTS-0.2.2 → AutoTS-0.2.3}/TODO.md

@@ -1,17 +1,24 @@
 # Basic Tenants
-* Ease of Use > Accuracy > Speed
+* Ease of Use > Accuracy > Speed (with speed more important with 'fast' selections)
+* The goal is to be able to run a horizontal ensemble prediction on 1,000 series/hour with a 'fast' selection, 10,000 series/hour with 'very fast'.
+* Availability of models which share information among series
 * All models should be probabilistic (upper/lower forecasts)
-* All models should be able to handle multiple parallel time series (even if just a `for` loop)
-* The expectation is series will largely be consistent in period, or at least up-sampled to regular intervals
+* All models should be able to handle multiple parallel time series
 * New transformations should be applicable to many datasets and models
 * New models need only be sometimes applicable
 * Fault tolerance: it is perfectly acceptable for model parameters to fail on some datasets, the higher level API will pass over and use others.
+* Missing data tolerance: large chunks of data can be missing and model will still produce reasonable results (although lower quality than if data is available)
 
-# Latest:
-* `grouping`/hierarchial reconciliation to GeneralTransformer
-* allow wide-style data as input
-* iterative imputer
-* allow a list of intervals to prediction_intervals in .predict()
+## Assumptions on Data
+* Series will largely be consistent in period, or at least up-sampled to regular intervals
+* The most recent data will generally be the most important
+* Forecasts are desired for the future immediately following the most recent data.
+
+# Latest
+* Added n_jobs parameters to pass through to joblib (although a joblib context manager is perhaps the best way)
+* Added joblib multiprocessing to ETS, GLM, and FBProphet
+* Fixed future warnings with pandas.DatetimeIndex.week and changes to statsmodels ETS
+* standardized source code formatting
 
 # Errors: 
 DynamicFactor holidays 	Exceptions 'numpy.ndarray' object has no attribute 'values'
@@ -24,7 +31,7 @@ VAR ValueError('Length of passed values is 4, index implies 9',)
 WindowRegression + KerasRNN + 1step + univariate = ValueError('Length mismatch: Expected axis has 54 elements, new values have 9 elements',)
 Is Template Eval Error: ValueError('array must not contain infs or NaNs',) related to Point to Probability HISTORIC QUANTILE?
 'Fake Date' doesn't work on entirely NaN series - ValueError('Found array with 0 sample(s) (shape=(0, 1)) while a minimum of 1 is required.',)
-
+ValueError: percentiles should all be in the interval [0, 1]. Try [-0.00089  0.01089] instead. .sample in Motif Simulation line 729 point_method == 'sample'
 
 ### Ignored Errors:
 xgboost poisson loss does not accept negatives
@@ -32,35 +39,59 @@ GluonTS not accepting quite a lot of frequencies
 KerasRNN errors due to parameters not working on all dataset
 Tensorflow GPU backend may crash on occasion.
 
-## General Tasks
+## Concerns
+* resource utilization at scale
+	* improve horizontal ensembling efficiency in particular
+* structure of General Transformer
+* overfitting on first train segment (train progressive subsets?)
+* End Users add their own models
+* improve starting templates (sorta best to wait until other things ironed out)
+* Improve documentation and usability of lower level code
+* better metrics, perhaps improve contour
+* better summarization of many time series into a few high-information time series as parallel or regressors
+* Ability to automatically add useful global information as regressors or parallel series
+* Built in GUI or Command Line tools
+
+
+## To-Do
+* Remove SP500 from AutoTS
+* ARIMA deprecation in Statsmodels
+* Migrate to-do to GitHub issues and project board
+	* GitHub badges in Readme, GitHub Actions flake8
+* Horizontal improvements
+	* don't run univariate models on all series, only on needed series
+	* generalizable from run on only a subset
+	* handle failure of a lower level model on some series (but not for dist variants)
+	* remove 'horizontal' sanity check run, takes too longer
+	* allow multiprocessing for each model of horizontal
+		* allow multiprocessing to models, pass through num_process + **kwargs
+* User friendly:
+	* clean up lower level
+	* make passing in own models easy
+		* clean up base model object
+* Make preprocessing templates more flexible...
+* Speed:
+	* Fast window regression only
+	* Fast MotifSimulation
+		* could memoization of pairwise comparisons be possible? (joblib)
+* Improve templates
+	* 'fake date' dataset with high diversity of series to train on
+
+
 * Profile slow parts of AutoTS on 1,000 series
 	* remove slow transformers unless parameter
 	* 'fast' option for RandomTransformations generator
-* constraint - TO TEMPLATE
-* coerce integer - TO HIGHER
-* ml model to generalize horizontal ensembles to many series after only trained on subset
-	* all be 'horizontal-max' and then have subset make smaller if desired
-	* have subsetting sample for diversity, not just random in this case
-	* handle failure of one of the models to predict
-* horizontal ensemble using prior selection of only one per class, and maybe also for speed
-	* cleanse similar models out first
-TemplateWizard catches errors right now
-Need to handle errors for ensemble lower level's but not for standalone modelsd
-Dicts instead of list of DFs
-Create model_id even if forecast fails
-Best3 to work on model failure
-	Make best 3 work on any given length of ensembles (more than 3, down to 1)
-	Rename from Best3
-	Add 'model_count' to parameters
-Horizontal to work on model failure
-	Use same process as to generalize to untested models
-Dist to fail on model failure
-HDist to fail on model failure
-
-# To-Do
+* have subsetting sample for diversity, not just random
+* cleanse similar models out first, before horizontal ensembling
+* Best3Ensemble
+	* Work on model failure
+	* Rename from Best3 to BestN
+	* Dicts instead of list of DFs
+	* Add 'model_count' to parameters
+
+* check models from M5 competition results
+* minmaxscaler as scoring for weighted Score generation
 * drop duplicates as function of TemplateEvalObject
-* speed up MotifSimulation
-* fake date dataset of many series to improve General Template
 * optimize randomtransform probabilities
 * improve test.py script for actual testing of many features
 * Add to template: Gluon, Motif, WindowRegression
@@ -95,11 +126,9 @@ HDist to fail on model failure
 	* Modify GluonStart if lots of NaN at start of that series
 	* GPU and CPU ctx
 * implement 'borrow' Genetic Recombination for ComponentAnalysis
-* Regressor to TensorflowSTS
 * Relative/Absolute Imports and reduce package reloading messages
 * Replace OrdinalEncoder with non-external code
 * 'Age' regressor as an option in addition to User/Holiday in ARIMA, etc.
-* Speed improvements, Profiling
 * Multiprocessing or Distributed options (Dask) for general greater speed
 * Improve usability on rarer frequenices (ie monthly data where some series start on 1st, others on 15th, etc.)
 * Figures: Add option to output figures of train/test + forecast, other performance figures
@@ -111,7 +140,6 @@ HDist to fail on model failure
 	* Code/documentation quality checkers
 * Ability to automatically add external datasets of parallel time series of global usability (ie from FRED or others)
 * make datetime input optional, just allow dataframes of numbers
-* Option to import either long or wide data
 * Infer column names for df_long to wide based on which is datetime, which is string, and which is numeric
 
 ### Links
@@ -153,7 +181,8 @@ HDist to fail on model failure
 	TPOT if it adds multioutput functionality
 	https://towardsdatascience.com/pyspark-forecasting-with-pandas-udf-and-fb-prophet-e9d70f86d802
 	Compressive Transformer
-	Reinforcement Learning
+	Reinforcement Learning with online forecasting?
+	
 
 #### New Transformations:
 	Sklearn iterative imputer 

{AutoTS-0.2.2 → AutoTS-0.2.3}/autots/__init__.py

@@ -13,11 +13,19 @@ from autots.evaluator.auto_ts import AutoTS
 from autots.tools.transform import GeneralTransformer
 from autots.tools.shaping import long_to_wide
 
-__version__ = '0.2.2'
+__version__ = '0.2.3'
 
 
-__all__ = ['load_daily','load_monthly', 'load_yearly', 'load_hourly', 'load_weekly',
-           'AutoTS', 'GeneralTransformer', 'long_to_wide']
+__all__ = [
+    'load_daily',
+    'load_monthly',
+    'load_yearly',
+    'load_hourly',
+    'load_weekly',
+    'AutoTS',
+    'GeneralTransformer',
+    'long_to_wide',
+]
 
 # import logging
 # logger = logging.getLogger(__name__)

{AutoTS-0.2.2 → AutoTS-0.2.3}/autots/datasets/__init__.py

@@ -7,4 +7,4 @@ from autots.datasets._base import load_yearly
 from autots.datasets._base import load_hourly
 from autots.datasets._base import load_weekly
 
-__all__ = ['load_daily', 'load_monthly', 'load_yearly', 'load_hourly', 'load_weekly']
+__all__ = ['load_daily', 'load_monthly', 'load_yearly', 'load_hourly', 'load_weekly']

{AutoTS-0.2.2 → AutoTS-0.2.3}/autots/datasets/_base.py

@@ -9,8 +9,9 @@ def load_daily():
     data_file_name = join(module_path, 'data', 'toy_daily.csv')
 
     df_long = pd.read_csv(data_file_name)
-    df_long['datetime'] = pd.to_datetime(df_long['datetime'],
-                                         infer_datetime_format=True)
+    df_long['datetime'] = pd.to_datetime(
+        df_long['datetime'], infer_datetime_format=True
+    )
 
     return df_long
 
@@ -19,9 +20,9 @@ def load_fred_monthly():
     """
     Federal Reserve of St. Louis.
     from autots.datasets.fred import get_fred_data
-    SeriesNameDict = {'GS10':'10-Year Treasury Constant Maturity Rate', 
-                              'MCOILWTICO':'Crude Oil West Texas Intermediate Cushing Oklahoma', 
-                              'CSUSHPISA': ' U.S. National Home Price Index', 
+    SeriesNameDict = {'GS10':'10-Year Treasury Constant Maturity Rate',
+                              'MCOILWTICO':'Crude Oil West Texas Intermediate Cushing Oklahoma',
+                              'CSUSHPISA': ' U.S. National Home Price Index',
                               'EXUSEU': 'US Euro Foreign Exchange Rate',
                               'EXCHUS': 'China US Foreign Exchange Rate',
                               'EXCAUS' : 'Canadian to US Dollar Exchange Rate Daily',
@@ -35,8 +36,9 @@ def load_fred_monthly():
     data_file_name = join(module_path, 'data', 'fred_monthly.zip')
 
     df_long = pd.read_csv(data_file_name, compression='zip')
-    df_long['datetime'] = pd.to_datetime(df_long['datetime'],
-                                         infer_datetime_format=True)
+    df_long['datetime'] = pd.to_datetime(
+        df_long['datetime'], infer_datetime_format=True
+    )
 
     return df_long
 
@@ -51,7 +53,7 @@ def load_fred_yearly():
     Federal Reserve of St. Louis.
     from autots.datasets.fred import get_fred_data
     SSeriesNameDict = {'GDPA':"Gross Domestic Product",
-                  'ACOILWTICO':'Crude Oil West Texas Intermediate Cushing Oklahoma', 
+                  'ACOILWTICO':'Crude Oil West Texas Intermediate Cushing Oklahoma',
                   'AEXUSEU': 'US Euro Foreign Exchange Rate',
                   'AEXCHUS': 'China US Foreign Exchange Rate',
                   'AEXCAUS' : 'Canadian to US Dollar Exchange Rate Daily',
@@ -69,8 +71,9 @@ def load_fred_yearly():
     data_file_name = join(module_path, 'data', 'fred_yearly.csv')
 
     df_long = pd.read_csv(data_file_name)
-    df_long['datetime'] = pd.to_datetime(df_long['datetime'],
-                                         infer_datetime_format=True)
+    df_long['datetime'] = pd.to_datetime(
+        df_long['datetime'], infer_datetime_format=True
+    )
 
     return df_long
 
@@ -82,15 +85,16 @@ def load_yearly():
 
 def load_traffic_hourly():
     """
-    From the MN DOT via the UCI data repository. 
+    From the MN DOT via the UCI data repository.
     Yes, Minnesota is the best state of the Union.
     """
     module_path = dirname(__file__)
     data_file_name = join(module_path, 'data', 'traffic_hourly.zip')
 
     df_long = pd.read_csv(data_file_name, compression='zip')
-    df_long['datetime'] = pd.to_datetime(df_long['datetime'],
-                                         infer_datetime_format=True)
+    df_long['datetime'] = pd.to_datetime(
+        df_long['datetime'], infer_datetime_format=True
+    )
 
     return df_long
 
@@ -106,8 +110,9 @@ def load_eia_weekly():
     data_file_name = join(module_path, 'data', 'eia_weekly.zip')
 
     df_long = pd.read_csv(data_file_name, compression='zip')
-    df_long['datetime'] = pd.to_datetime(df_long['datetime'],
-                                         infer_datetime_format=True)
+    df_long['datetime'] = pd.to_datetime(
+        df_long['datetime'], infer_datetime_format=True
+    )
     return df_long
 
 

AutoTS-0.2.3/autots/datasets/fred.py

@@ -0,0 +1,72 @@
+"""
+FRED (Federal Reserve Economic Data) Data Import
+
+requires API key from FRED
+and pip install fredapi
+"""
+import pandas as pd
+
+try:
+    from fredapi import Fred
+except Exception:  # except ImportError
+    _has_fred = False
+else:
+    _has_fred = True
+
+
+def get_fred_data(fredkey: str, SeriesNameDict: dict = {'SeriesID': 'SeriesName'}):
+    """
+    Imports Data from Federal Reserve
+
+    args:
+        fredkey - an API key from FRED
+
+        SeriesNameDict, pairs of FRED Series IDs and Series Names
+            Series id must match Fred IDs, but name can be anything
+            if default is use, several default samples are returned
+    """
+    if not _has_fred:
+        raise ImportError("Package fredapi is required")
+
+    fred = Fred(api_key=fredkey)
+
+    if SeriesNameDict == {'SeriesID': 'SeriesName'}:
+        SeriesNameDict = {
+            'T10Y2Y': '10 Year Treasury Constant Maturity Minus 2 Year Treasury Constant Maturity',
+            'DGS10': '10 Year Treasury Constant Maturity Rate',
+            'DCOILWTICO': 'Crude Oil West Texas Intermediate Cushing Oklahoma',
+            'SP500': 'S&P 500',
+            'DEXUSEU': 'US Euro Foreign Exchange Rate',
+            'DEXCHUS': 'China US Foreign Exchange Rate',
+            'DEXCAUS': 'Canadian to US Dollar Exchange Rate Daily',
+            'VIXCLS': 'CBOE Volatility Index: VIX',  # this is a more irregular series
+            'T10YIE': '10 Year Breakeven Inflation Rate',
+            'USEPUINDXD': 'Economic Policy Uncertainty Index for United States',  # also very irregular
+        }
+
+    series_desired = list(SeriesNameDict.keys())
+
+    fred_timeseries = pd.DataFrame(
+        columns=['date', 'value', 'series_id', 'series_name']
+    )
+
+    for series in series_desired:
+        data = fred.get_series(series)
+        try:
+            series_name = SeriesNameDict[series]
+        except Exception:
+            series_name = series
+        data_df = pd.DataFrame(
+            {
+                'date': data.index,
+                'value': data,
+                'series_id': series,
+                'series_name': series_name,
+            }
+        )
+        data_df.reset_index(drop=True, inplace=True)
+        fred_timeseries = pd.concat(
+            [fred_timeseries, data_df], axis=0, ignore_index=True
+        )
+
+    return fred_timeseries

{AutoTS-0.2.2 → AutoTS-0.2.3}/autots/evaluator/__init__.py

@@ -1,4 +1,3 @@
 """
 Model Evaluators
 """
-