ulfsky.github.io

This repository contains projects that have been done for Practicum Yandex100 training program.

Project notebook

Data

The data is stored in 2 tables.

The dataset telecom_dataset_us.csv contains the following columns:

• direction — call direction (out for outgoing, in for incoming)
  
• internal — whether the call was internal (between a client’s operators)
  
• operator_id — operator identifier
  
• is_missed_call — whether the call was missed
  
• date — date the statistics were retrieved
  
• user_id — client account ID
  
• calls_count — number of calls
  
• call_duration — call duration (excluding waiting time)
  
• total_call_duration — call duration (including waiting time)
  

The dataset telecom_clients_us.csv has the following columns:

• tariff_plan — client’s current plan
  
• date_start — client’s registration date
  

---

Goal:

to study effectivness of phone operators for virtual telephony service CallMeMaybe.
To define ineffective operators we will study the following features:

• number of missed incoming calls (internal and external),
  
• waiting time for incoming calls,
  
• number of outgoing calls,
  
• number of internal calls between operators.
  

Also we will study these features for every tariff plan the company has (A, B, C) and define ineffective operators for every of this plan.

We will test the hypothesis if the difference in operators performance for different tariff plans is statistically significant.
Finally we will draw basic conclusions and develop some recommendations on how to improve operators performance.
__

Libraries used:

• pandas,
• IPython.display,
• scipy,
• matplotlib,
• numpy,
• functools,
• stats,
• seaborn,
• math,
• plotly