class ModuleEngagementSummaryDataTaskMapTest(MapperTestMixin, TestCase):
"""Base class for test analysis of student engagement summaries"""
task_class = ModuleEngagementSummaryDataTask
input_record = ModuleEngagementRecord(course_id='foo/bar/baz',
username='******',
date=datetime.date(2015, 11, 1),
entity_type='problem',
entity_id='problem-id',
event='attempted',
count=1)
def test_invalid_input_types(self):
input_as_strings = list(self.input_record.to_string_tuple())
input_as_strings[-1] = 'foo'
with self.assertRaises(ValueError):
tuple(self.task.mapper('\t'.join(input_as_strings) + '\n'))
def test_not_enough_input_columns(self):
input_as_strings = list(self.input_record.to_string_tuple())[:-1]
with self.assertRaises(ValueError):
tuple(self.task.mapper('\t'.join(input_as_strings) + '\n'))
def test_single_output(self):
tsv_line = self.input_record.to_separated_values()
self.assert_single_map_output(tsv_line, ('foo/bar/baz', 'foouser'),
tsv_line)
class ModuleEngagementSummaryDataTaskReducerTest(ReducerTestMixin, TestCase):
"""Base class for test analysis of student engagement summaries"""
task_class = ModuleEngagementSummaryDataTask
output_record_type = ModuleEngagementSummaryRecord
input_record = ModuleEngagementRecord(course_id='foo/bar/baz',
username='******',
date=datetime.date(2014, 03, 26),
entity_type='problem',
entity_id='problem-id',
event='attempted',
count=1)
def setUp(self):
super(ModuleEngagementSummaryDataTaskReducerTest, self).setUp()
self.reduce_key = (self.COURSE_ID, 'test_user')
self.video_play_record = self.input_record.replace(
entity_type='video',
entity_id='foox-video1',
event='viewed',
)
self.forum_record = self.input_record.replace(entity_type='discussion',
entity_id='forum0',
event='contributed')
def test_output_format(self):
self._check_output_complete_tuple(
[self.input_record.to_separated_values()], ((
'foo/bar/baz',
'test_user',
'2014-03-25',
'2014-04-01',
'1',
'1',
'0',
'inf',
'0',
'0',
'1',
), ))
def test_multiple_problem_attempts_same_problem(self):
self._check_output_by_record_field(
[
self.input_record.to_separated_values(),
self.input_record.replace(
date=datetime.date(2014, 03, 27)).to_separated_values()
], {
'problem_attempts': '2',
'problems_attempted': '1',
'problem_attempts_per_completed': 'inf',
'days_active': '2'
})
def test_multiple_problem_attempts_different_problem(self):
self._check_output_by_record_field(
[
self.input_record.to_separated_values(),
self.input_record.replace(
entity_id='problem-id-2').to_separated_values()
], {
'problem_attempts': '2',
'problems_attempted': '2',
'problem_attempts_per_completed': 'inf',
'days_active': '1'
})
def test_correct_problem_attempt(self):
self._check_output_by_record_field(
[
self.input_record.to_separated_values(),
self.input_record.replace(
event='completed').to_separated_values()
], {
'problem_attempts': '1',
'problems_attempted': '1',
'problems_completed': '1',
'problem_attempts_per_completed': str(1.0),
'days_active': '1'
})
def test_correct_problem_multiple_attempts(self):
self._check_output_by_record_field(
[
self.input_record.replace(count=4).to_separated_values(),
self.input_record.replace(
event='completed').to_separated_values()
], {
'problem_attempts': '4',
'problems_attempted': '1',
'problems_completed': '1',
'problem_attempts_per_completed': str(4.0),
'days_active': '1'
})
def test_multiple_correct_problems(self):
self._check_output_by_record_field(
[
self.input_record.replace(count=4).to_separated_values(),
self.input_record.replace(
event='completed').to_separated_values(),
self.input_record.replace(
date=datetime.date(
2014, 03, 27), entity_id='p2').to_separated_values(),
self.input_record.replace(
date=datetime.date(2014, 03, 27),
entity_id='p2',
event='completed',
).to_separated_values(),
], {
'problem_attempts': '5',
'problems_attempted': '2',
'problems_completed': '2',
'problem_attempts_per_completed': str(2.5),
'days_active': '2'
})
def test_video_viewed(self):
self._check_output_by_record_field(
[
self.video_play_record.to_separated_values(),
], {
'problem_attempts': '0',
'problems_attempted': '0',
'problems_completed': '0',
'problem_attempts_per_completed': 'inf',
'videos_viewed': '1',
'days_active': '1'
})
def test_multiple_videos_viewed(self):
self._check_output_by_record_field([
self.video_play_record.to_separated_values(),
self.video_play_record.replace(
entity_id='1').to_separated_values(),
], {'videos_viewed': '2'})
def test_multiple_forum_contributions(self):
self._check_output_by_record_field([
self.forum_record.to_separated_values(),
self.forum_record.replace(entity_id='1',
count=2).to_separated_values(),
], {'discussion_contributions': '3'})
def requires(self):
kwargs = {
'warehouse_path': self.warehouse_path,
'overwrite': True,
'schema': self.schema,
'credentials': self.credentials,
'read_timeout': self.read_timeout,
'marker_schema': self.marker_schema,
}
yield (
LoadHiveTableToVertica(
table_name='course_activity',
sql_schema=CourseActivityRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='course_enrollment_daily',
sql_schema=EnrollmentDailyRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='course_enrollment_birth_year_daily',
sql_schema=EnrollmentByBirthYearRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='course_enrollment_education_level_daily',
sql_schema=EnrollmentByEducationLevelRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='course_enrollment_gender_daily',
sql_schema=EnrollmentByGenderRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='course_enrollment_mode_daily',
sql_schema=EnrollmentByModeRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='course_meta_summary_enrollment',
sql_schema=CourseSummaryEnrollmentRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='course_program_metadata',
sql_schema=CourseProgramMetadataRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='course_enrollment_location_current',
load_from_latest_partition=False,
sql_schema=LastCountryPerCourseRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='module_engagement',
sql_schema=ModuleEngagementRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='module_engagement_metric_ranges',
sql_schema=ModuleEngagementSummaryMetricRangeRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='video_timeline',
sql_schema=VideoTimelineRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='video',
sql_schema=VideoSegmentSummaryRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='last_country_of_user_id',
sql_schema=LastCountryOfUserRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='enterprise_enrollment',
sql_schema=EnterpriseEnrollmentRecord.get_sql_schema(),
**kwargs
),
LoadHiveTableToVertica(
table_name='enterprise_user',
sql_schema=EnterpriseUserRecord.get_sql_schema(),
**kwargs
)
)
def requires(self):
kwargs = {
'warehouse_path': self.warehouse_path,
'overwrite': True,
'schema': self.schema,
'credentials': self.credentials,
'read_timeout': self.read_timeout,
'marker_schema': self.marker_schema,
}
yield (
LoadHiveTableToVertica(
table_name='course_activity',
sql_schema=CourseActivityRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='course_enrollment_daily',
sql_schema=EnrollmentDailyRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='course_enrollment_birth_year_daily',
sql_schema=EnrollmentByBirthYearRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='course_enrollment_education_level_daily',
sql_schema=EnrollmentByEducationLevelRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='course_enrollment_gender_daily',
sql_schema=EnrollmentByGenderRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='course_enrollment_mode_daily',
sql_schema=EnrollmentByModeRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='course_meta_summary_enrollment',
sql_schema=CourseSummaryEnrollmentRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='course_program_metadata',
sql_schema=CourseProgramMetadataRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='course_enrollment_location_current',
load_from_latest_partition=False,
sql_schema=LastCountryPerCourseRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='module_engagement',
sql_schema=ModuleEngagementRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='module_engagement_metric_ranges',
sql_schema=ModuleEngagementSummaryMetricRangeRecord.
get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='video_timeline',
sql_schema=VideoTimelineRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='video',
sql_schema=VideoSegmentSummaryRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='last_country_of_user_id',
sql_schema=LastCountryOfUserRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='enterprise_enrollment',
sql_schema=EnterpriseEnrollmentRecord.get_sql_schema(),
**kwargs),
LoadHiveTableToVertica(
table_name='enterprise_user',
sql_schema=EnterpriseUserRecord.get_sql_schema(),
**kwargs))