def __init__(self):
feature_cols = [
sql.Column('sixth_read_pl', VARCHAR),
sql.Column('sixth_math_pl', VARCHAR),
sql.Column('sixth_write_pl', VARCHAR),
sql.Column('sixth_ctz_pl', VARCHAR),
sql.Column('sixth_science_pl', VARCHAR),
sql.Column('seventh_read_pl', VARCHAR),
sql.Column('seventh_math_pl', VARCHAR),
sql.Column('seventh_write_pl', VARCHAR),
sql.Column('eighth_read_pl', VARCHAR),
sql.Column('eighth_math_pl', VARCHAR),
sql.Column('eighth_science_pl', VARCHAR),
sql.Column('eighth_socstudies_pl', VARCHAR),
]
pre_features_processor = CompositeFeatureProcessor([
# OAAOGTProcessor(), # adding this reduces testing perf
CategoricalFeatureProcessor(
column_list=oaaogt_features_config.categorical_columns)
])
post_features_processor = CompositeFeatureProcessor(
[ImputeNullProcessor(fill_unspecified=ImputeBy.MEAN)])
super(OAAOGTFeatures, self).__init__(
table_name='oaaogt_features',
feature_cols=feature_cols,
categorical_cols=[],
post_features_processor=post_features_processor,
pre_features_processor=pre_features_processor,
high_school_features=False,
)
def get_default_feature_processors(train_stats=None):
train_stats = [None] * 2 if train_stats is None else train_stats
return CompositeFeatureProcessor([
ImputeNullProcessor(train_stats=train_stats[0],
fill_unspecified=ImputeBy.MEAN,
col_flag_set=False),
StandardizeProcessor(train_stats=train_stats[1])
])
def __init__(self):
joined_absences = self.get_joined_absences()
super(AbsenceDescFeatures, self).__init__(
table_name=inflection.underscore(AbsenceDescFeatures.__name__),
categorical_cols=[],
post_features_processor=CompositeFeatureProcessor(
[ImputeNullProcessor(fill_unspecified=0, col_flag_set=False)]),
data_table=joined_absences,
blocking_col=joined_absences.c.absence_desc,
)
def __init__(self):
feature_cols = [
sql.Column('discipline_incident_rate', FLOAT),
sql.Column('discipline_incident_rate_perc', FLOAT)
]
super(DisciplineIncidentRateFeatures, self).__init__(
table_name=inflection.underscore(
DisciplineIncidentRateFeatures.__name__),
feature_cols=feature_cols,
categorical_cols=[],
post_features_processor=CompositeFeatureProcessor(
[ImputeNullProcessor(col_val_dict=config.fill_values)]))
def __init__(self):
feature_cols = [
sql.Column('absence_rate',FLOAT),
sql.Column('unexcused_absence_rate',FLOAT),
sql.Column('absence_rate_perc',FLOAT)
]
super(AbsenceFeatures, self).__init__(
table_name=inflection.underscore(AbsenceFeatures.__name__), # converts AbsenceFeatures to 'absence_features'
feature_cols=feature_cols,
categorical_cols=[],
post_features_processor=CompositeFeatureProcessor([
ImputeNullProcessor(
col_val_dict= absence_features_config.fill_values
)
])
)
def __init__(self):
cols = [
sql.Column('pivot_ms_grade', INT),
sql.Column('pivot_ms_avg_gpa', FLOAT),
]
feature_processor = CompositeFeatureProcessor(
[ImputeNullProcessor(fill_unspecified=ImputeBy.MEAN)])
super(MiddleSchoolFeatures, self).__init__(
table_name='middle_school_features',
feature_cols=cols,
categorical_cols=[],
post_features_processor=feature_processor,
pre_features_processor=PivotProcessor(index='student_lookup',
columns='pivot_ms_grade',
values='pivot_ms_avg_gpa'),
high_school_features=False)
def __init__(self):
all_inv = db_tables.clean_intervention_table
index_cols_dict = {
'student_lookup':
all_inv.c.student_lookup,
'school_year':
sql.cast(
db_func.substr(all_inv.c.school_year,
db_func.length(all_inv.c.school_year) - 3, 4),
sql.INT).label('school_year'),
'grade':
all_inv.c.grade
}
super(InvFeatures, self).__init__(
table_name=inflection.underscore(InvFeatures.__name__),
categorical_cols=inv_features_config.categorical_columns,
post_features_processor=CompositeFeatureProcessor(
[ImputeNullProcessor(fill_unspecified=0)]),
data_table=all_inv,
blocking_col=all_inv.c.inv_group,
index_cols_dict=index_cols_dict)
def __init__(self):
cols = [
sql.Column('num_classes', INT),
sql.Column('hs_avg_gpa', FLOAT),
sql.Column('last_year_gpa', FLOAT),
sql.Column('pre_2_year_gpa', FLOAT),
sql.Column('pre_3_year_gpa', FLOAT),
sql.Column('overall_percentile', FLOAT),
sql.Column('district_percentile', FLOAT),
sql.Column('school_percentile', FLOAT),
]
feature_processor = CompositeFeatureProcessor([
ImputeNullProcessor(
col_val_dict=academic_features_config.fill_values,
fill_unspecified=False
)
])
super(AcademicFeatures, self).__init__(
table_name='academic_features',
feature_cols=cols,
categorical_cols=[],
post_features_processor=feature_processor
)
def __init__(self):
cols = [
sql.Column('gender', VARCHAR),
sql.Column('ethnicity', VARCHAR),
sql.Column('school_name', VARCHAR),
sql.Column('district', VARCHAR),
sql.Column('disability', VARCHAR),
sql.Column('disadvantagement', VARCHAR),
sql.Column('economic_disadvantagement', INT),
sql.Column('academic_disadvantagement', INT),
sql.Column('limited_english', VARCHAR),
sql.Column('discipline_incidents', INT),
sql.Column('pre_1_year_discipline_incidents', INT),
sql.Column('pre_2_year_discipline_incidents', INT),
sql.Column('days_absent', FLOAT),
sql.Column('pre_1_year_days_absent', FLOAT),
sql.Column('pre_2_year_days_absent', FLOAT),
sql.Column('age', INT),
sql.Column('num_transfers', INT),
sql.Column('cumul_discipline_incidents', INT),
]
feature_processor = CompositeFeatureProcessor([
ReplaceNullishProcessor(
column_list=snapshot_features_config.replace_nullish_columns),
ImputeNullProcessor(
col_val_dict=snapshot_features_config.fill_values,
col_flag_set=snapshot_features_config.impute_flag_columns),
CategoricalFeatureProcessor(
column_list=snapshot_features_config.categorical_columns)
])
super(SnapshotFeatures, self).__init__(
table_name='snapshot_features',
feature_cols=cols,
categorical_cols=snapshot_features_config.categorical_columns,
post_features_processor=feature_processor)