canvs-toolbox-canvsUser 0.1.2

A package for custom canvs analytics

Canvs Toolbox Package

Date Formatting

"MM/DD/YY"

General

from canvs_toolbox import general as gen

• gen.consolidate_data(file_path, file_type='csv')

API Tools

Canvs TV

from canvs_toolbox.api import tv as tvAPI

• tvAPI.twitter_daily_export(api_key, data_mode, start_date, end_date)
• tvAPI.twitter_emotional_authors(api_key, series_id, start_date, end_date)
• tvAPI.airings_backfill(api_key, data_mode, start_date, end_date)
• tvAPI.facebook_backfill(api_key, data_mode, start_date, end_date)

Canvs Watch

from canvs_toolbox.api import watch as watchAPI

• watchAPI.post_backfill(api_key, data_mode, start_date, end_date)
• watchAPI.series_backfill(api_key, data_mode, start_date, end_date)

Canvs Social

from canvs_toolbox.api import social as socialAPI

• socialAPI.get_facebook_posts(api_key, fb_id, org_id, start_date, end_date, query_increment=None)
• socialAPI.get_page_collection(api_key, org_id, start_date, end_date, fb_pages, query_increment=None)

Analytics Tools

Canvs TV

from canvs_toolbox.analytics import tv as tvAnalytics

Audience Overlap Analysis

• implementation: tvAnalytics.audience_overlap_analysis(directory)
• input directory should contain audience csv files from calling tvAPI.twitter_emotional_authors()
• analysis will find the overlapping audiences across those csv files
• for best results, rename the csv files to desired series names (e.g. showA, showB)

Audience Erosion Analysis

• implementation: tvAnalytics.audience_erosion_analysis(filename)
• input file should be a single audience csv file from calling tvAPI.twitter_emotional_authors()
• creates an episode-over-episode drop-off analysis

Emotional Fingerprinting Analysis

• implementation: emotional_fingerprinting_analysis(source, filename, format)
• input file should be either from a direct Explore Programs export from the Canvs App or from any of the API exports except twitter_emotional_authors
• computes an emotional similarity score for all possible combinations of passed-in content
• can choose to either return a stacked view of all pairings and their scores (format = 'stacked) or a matrix view containing similarity scores at content intersections (format = 'matrix')