eandualem / telecom_analytics

Prior to investing in a new company, a thorough analysis of the data behind the company, and above all; the identification of opportunities to boost profitability is essential. The main goal of this project is to analyze Telco's data to determine whether it is worth buying or selling.

Telecom Analytics

Table of content

• Telecom Analytics
  • Overview
  • Requirements
  • Install
  • Data
  • Features
    • Data Exploration
      • 1.0_preprocessing.ipynb:
      • 1.01_outliers.ipynb
      • 1.1_overview:
      • 1.2_data_analysis:
    • User Engagement Analysis
    • User Experience Analysis
    • User Satisfaction Analysis
    • Scripts
    • Test
    • Travis CI
    • Dashboard

Overview

This repository analyses users’ experience, engagement, and satisfaction to check for growth opportunities in TellCo, a mobile service provider.

Requirements

Python 3.5 and above, Pip and MYSQL The visualization are made using plotly. I am showing only static images in the notebooks bc interactive plots are not visible in GitHub. If you set, interactive true the plots will be interactive. eg:

• hist(df): static image
• hist(df, interactive=True): interactive plot

Install

git clone https://github.com/eandualem/telecom_analytics.git
cd telecom_analytics
pip install -r requirements.txt

Data

• The data used in the project is generated automatically by TellCo systems.
• The data is here - extracted from a month of aggregated data on xDR.
• The features described can be found here.

Features

Data Exploration

• Data exploration is done in 4 notebooks

1.0_preprocessing.ipynb:

• Here I have cleaned the original data by
  • renaming column labels
  • removing duplicates
  • handle null values
  • convert data types

1.01_outliers.ipynb

• Here I have calculated outliers in each column and their percentages
• If the percentage is small i have dropped the rows with outliers
• For columns with a larger percentage of outliers, I have replaced them with values using the IQR.
• Then I have selected columns for further analysis and saved them in clean_data.csv.

1.1_overview:

• I have done task 1.1 here.

1.2_data_analysis:

• I have done task 1.2 here.

User Engagement Analysis

• User engagement analysis is done in 2.0_engagement_analysis.ipynb.

User Experience Analysis

• User experience analysis is done in 3.0_experience_analytics.ipynb.

User Satisfaction Analysis

• User satisfaction analysis is done in 4.0_satisfaction_analysis.ipynb.

Scripts

• All the scripts used by the notebooks are inside the scripts folder.

Test

• Tests for the scripts are inside the tests folder.

Travis CI

• The file .travis.yml contains the configuration for Travis.

Dashboard

• The code for the dashboard is inside streamlit folder
• I have built Docker image that can be built and run.
• Its hosted on heroku
• link: https://ancient-coast-11237.herokuapp.com