From: Christophe Guyeux Date: Mon, 17 Feb 2020 14:03:29 +0000 (+0100) Subject: Refactoring, and categorical / numerical / mixed NaN values are now X-Git-Url: https://bilbo.iut-bm.univ-fcomte.fr/and/gitweb/predictops.git/commitdiff_plain/20a117b07643f7b3ef305d1e7a6f62f05e33698e?ds=inline;hp=4b6d71d96bb92791cc31640e5f30378ae6fe63e4 Refactoring, and categorical / numerical / mixed NaN values are now filled accordingly. --- diff --git a/README.md b/README.md index e224f20..e2030e7 100644 --- a/README.md +++ b/README.md @@ -10,7 +10,7 @@ Initialisation de l'environnement de travail `python -m venv ~/.venvs/predictops` -activer l'environnement : +- activer l'environnement : `source ~/.venvs/predictops/bin/activate` @@ -42,7 +42,10 @@ Tout se passe dans le répertoire features 2. Détailler le traitement de chaque famille de feature dans le cfg associé (feature_ephemeris.cfg, feature_meteo.cfg, etc.), en accord avec les fichiers -csv associés dans le répertoire features. +csv associés dans le répertoire features. Dans ces derniers, le type spécifie +si la variable est numérique (1), qualitative (2), ou si elle peut être consi- +dérée comme de l'un ou l'autre type (3), comme le jour dans l'année. + Exécution diff --git a/config/features/ephemeris_features.csv b/config/features/ephemeris_features.csv index 72060ef..4b75f8a 100644 --- a/config/features/ephemeris_features.csv +++ b/config/features/ephemeris_features.csv @@ -1,7 +1,7 @@ name,type hour,3 dayInWeek,3 -dayInMonth,3 +dayInMonth,2 dayInYear,3 weekInYear,3 month,3 diff --git a/config/feature_ephemeris.cfg b/config/features/feature_ephemeris.cfg similarity index 50% rename from config/feature_ephemeris.cfg rename to config/features/feature_ephemeris.cfg index 6b37dcf..ddd9f8b 100644 --- a/config/feature_ephemeris.cfg +++ b/config/features/feature_ephemeris.cfg @@ -7,8 +7,20 @@ weekInYear = True month = True year = True -[HOUR] +[hour] +numerical = False + +[dayInWeek] +numerical = False + +[dayInYear] +numerical = False + +[weekInYear] +numerical = False + +[month] numerical = True -[YEAR] +[year] numerical = True \ No newline at end of file diff --git a/config/feature_meteo.cfg b/config/features/feature_meteo.cfg similarity index 100% rename from config/feature_meteo.cfg rename to config/features/feature_meteo.cfg diff --git a/config/learn.cfg b/config/learn.cfg index bbd3557..9475c78 100644 --- a/config/learn.cfg +++ b/config/learn.cfg @@ -10,8 +10,8 @@ ephemeris = True [FEATURE_CONFIG] -meteofrance = (Path.cwd() / 'config') / 'feature_meteo.cfg' -ephemeris = (Path.cwd() / 'config') / 'feature_ephemeris.cfg' +meteofrance = (Path.cwd() / 'config') / 'features' / 'feature_meteo.cfg' +ephemeris = (Path.cwd() / 'config') / 'features' / 'feature_ephemeris.cfg' [PREPROCESSING] diff --git a/predictops/engine.py b/predictops/engine.py index 2ec62df..8ba5043 100644 --- a/predictops/engine.py +++ b/predictops/engine.py @@ -5,8 +5,8 @@ from logging.config import fileConfig from pathlib import Path from shutil import rmtree -from predictops.source.ephemeris import Ephemeris -from predictops.source.meteofrance import MeteoFrance +from .source.ephemeris import Ephemeris +from .source.meteofrance import MeteoFrance fileConfig((Path.cwd() / 'config') / 'logging.cfg') logger = getLogger() diff --git a/predictops/learn/preprocessing.py b/predictops/learn/preprocessing.py index 49d7ef8..939a7fa 100644 --- a/predictops/learn/preprocessing.py +++ b/predictops/learn/preprocessing.py @@ -48,16 +48,23 @@ class Preprocessing: else: self._features = set(chain.from_iterable([tuple(u.keys()) for u in [*dict_features.values()]])) - csv_files = Path.cwd() / 'config' / 'features' - self._features = {feat : None for feat in self._features} - for csv_file in listdir(csv_files): - with open(csv_files / csv_file, "r") as f: - reader = DictReader(f, delimiter=',') - for row in reader: - if row['name'] in self._features: - self._features[row['name']] = row['type'] - print(self._features) - exit() + feature_files = Path.cwd() / 'config' / 'features' + self._features = {feat : {'numerical': False} for feat in self._features} + for feature_file in listdir(feature_files): + if feature_file.endswith('csv'): + with open(feature_files / feature_file , "r") as f: + reader = DictReader(f, delimiter=',') + typed_names = {row['name']: row['type'] for row in reader} + for feature in self._features: + if feature.split('_')[0] in typed_names: + self._features[feature]['type'] = int(typed_names[feature.split('_')[0]]) + elif feature_file.endswith('cfg'): + config = ConfigParser() + config.read(feature_files / feature_file) + for section in config: + if config.has_option(section, 'numerical'): + self._features[section]['numerical'] = config[section].getboolean('numerical') + @property @@ -137,21 +144,37 @@ class Preprocessing: logger.info("Creating feature dataframe from feature dictionary") self._dataframe = pd.DataFrame.from_dict(self.full_dict, orient='index') - logger.info("Filling NaN values in the feature dataframe") - + logger.info("Filling NaN numerical values in the feature dataframe") + # We interpolate (linearly or with splines) only numerical columns + numerical_columns = [k for k in self._features if self._features[k]['type'] == 1 + or (self._features[k]['type'] == 3 and self._features[k]['numerical'])] + # The interpolation if self._config['PREPROCESSING']['fill_method'] == 'propagate': - self._dataframe = self._dataframe.fillna(method='ffill') + self._dataframe[numerical_columns] =\ + self._dataframe[numerical_columns].fillna(method='ffill') elif self._config['PREPROCESSING']['fill_method'] == 'linear': - self._dataframe = self._dataframe.interpolate() + self._dataframe[numerical_columns] =\ + self._dataframe[numerical_columns].interpolate() elif self._config['PREPROCESSING']['fill_method'] == 'spline': - self._dataframe = self._dataframe.interpolate(method='spline', - order=self._config['PREPROCESSING'].getint('order')) + self._dataframe[numerical_columns] =\ + self._dataframe[numerical_columns].interpolate(method='spline', + order=self._config['PREPROCESSING'].getint('order')) + + # For the categorical columns, NaN values are filled by duplicating + # the last known value (forward fill method) + logger.info("Filling NaN categorical values in the feature dataframe") + categorical_columns = [k for k in self._features if self._features[k]['type'] == 2 + or (self._features[k]['type'] == 3 and not self._features[k]['numerical'])] + self._dataframe[categorical_columns] =\ + self._dataframe[categorical_columns].fillna(method='ffill') # Uncomment this line to fill NaN values at the beginning of the # dataframe. This may not be a good idea, especially for features # that are available only for recent years, e.g., air quality #self._dataframe = self._dataframe.fillna(method='bfill') + # Dropping rows that are not related to our datetime window (start/ + # step / end) self._dataframe = self._dataframe.drop([k.to_pydatetime() for k in self._dataframe.T if k not in self._datetimes]) diff --git a/predictops/source/source.py b/predictops/source/source.py index 8e68716..70f24da 100644 --- a/predictops/source/source.py +++ b/predictops/source/source.py @@ -1,3 +1,4 @@ +from configparser import ConfigParser from csv import DictReader from logging import getLogger from logging.config import fileConfig @@ -11,14 +12,40 @@ logger = getLogger() class Source: def __init__(self): ''' - Check if the same feature name is used in two different feature sources + Check if the same feature name is used in two different feature sources, + and if the sources of type 3 (being both categorical and numerical) have + a specified type in the feature_...cfg file ''' logger.info('Check for redondant feature names') - csv_files = Path.cwd() / 'config' / 'features' + feature_files = Path.cwd() / 'config' / 'features' list_of_names = [] - for csv_file in listdir(csv_files): - with open(csv_files / csv_file, "r") as f: - reader = DictReader(f, delimiter=',') - list_of_names.extend([row['name'] for row in reader]) + for file_name in listdir(feature_files ): + if file_name.endswith('csv'): + with open(feature_files / file_name, "r") as f: + reader = DictReader(f, delimiter=',') + list_of_names.extend([row['name'] for row in reader]) + if len(list_of_names) != len(set(list_of_names)): - raise ValueError("At least two features have the same name") \ No newline at end of file + raise ValueError("At least two features have the same name") + + logger.info('Check for specified feature types') + names_of_mixed_types = [] + for file_name in listdir(feature_files): + if file_name.endswith('csv'): + with open(feature_files / file_name, "r") as f: + reader = DictReader(f, delimiter=',') + names_of_mixed_types.extend([row['name'] for row in reader + if row['type'] == '3']) + + cfg_names_of_mixed_types = [] + for file_name in listdir(feature_files): + if file_name.endswith('cfg'): + config = ConfigParser() + config.read(feature_files / file_name) + for section in config: + if config.has_option(section, 'numerical'): + cfg_names_of_mixed_types.append(section) + + if sorted(names_of_mixed_types) != sorted(cfg_names_of_mixed_types): + raise ValueError(f"Problem with features of mixed types: " + f"{set(names_of_mixed_types).symmetric_difference(cfg_names_of_mixed_types)}")