An initial implementation of the subscriber profiles using R6 approach of R language.

According to some benchmarks, R6 provides the most flexible way of data structures implementation using object-oriented techniques. Among those of R language (S4, R5, R6), it's R6 objects take roughly as much memory as the plain S3 objects (for simplicity, plain R functions), though they call for some tweaking too. These data structures and functions have been wrapped into a plain R package. As a result, a number of things intended to make the developer life to be easier (uncluding that boring setwd(...), install.packages(...), library(...) routine), are implemented right out of the box.

Please, take notice I have not included the original data because of privacy concerns.

Originally, this package was implemented to cover needs of a cellular provider. So the subscriber profile includes the following fields (self-explainable):

Initial info

hash_number_A, billing_tariff.plan, billing_arpu, billing_usage, billing_top.up,

Service usage

service.usage_renewal,,, service.usage_last.update,

Predictive model

pred.model_churn.score, pred.model_mnp.out, pred.model_change.tariff, pred.model_winback,

Social data

social.data_age, social.data_sex,, social.data_marriage, social.data_household, social.data_wager,

Technical stream

tech.stream_out.of.service, tech.stream_drop.calls, tech.stream_data.speed, tech.stream_location,

Value management

value.mngt_life.stage, value.mngt_value.over.time,, value.mngt_margin.optimisation,

Web data

web.data_cookie, web.data_browsing.history, web.data_adblock, web.data_popular.url, web.data_click.rate,


device_type, device_price, device_applications, device_last.update, hash_tariff , event , event_sub , network_service_direction, event_start_date , LAT , LON , cost, hash_number_B , number_B_category, call_duration_minutes, data_volume_mb, hash_accum_code, interest_1, interest_2, interest_3, interest_4, interest_5

av1611/subscriber documentation built on May 16, 2019, 6:57 p.m.