Network Performance Analysis in Bangladesh

Network Performance Analysis in Bangladesh#

Data Description#

This analysis uses Ookla Speedtest connectivity data, which provides user-generated network performance information. The dataset captures:

  • Download and upload speeds

  • Number of tests performed

  • Geographic location (district and region level)

  • Timestamps of tests

Data Characteristics#

  • The data is based on user-initiated speed tests performed on fixed network

  • Tests can be conducted on both fixed and broadband internet connections

  • Key metrics include:

    • Number of users conducting tests

    • Download speed (Mbps)

    • Upload speed (Mbps)

Methodology#

The analysis employs a multi-step approach to assess network performance:

  1. Baseline Establishment

    • For each district, calculate the average number of daily users conducting tests during July 1-15, 2024

    • This baseline period represents normal network testing activity before the shutdown event

  2. Data Validation and Filtering

    • Calculate daily test volumes by district

    • Compare against district-specific baselines

    • Apply 90% threshold filter:

      • If daily tests fall below 90% of the baseline average

      • Set corresponding speed measurements (download/upload) to zero

      • This addresses potential reliability issues during disruption periods

  3. Performance Metrics Analysis

    • Track three key indicators:

      • Number of tests conducted

      • Download speeds (Mbps)

      • Upload speeds (Mbps)

    • Calculate percentage changes from baseline period

    • Analyze patterns before, during, and after the shutdown period (July 18-23)

  4. Geographic Aggregation

    • Primary analysis at district level

    • Regional aggregation for broader patterns

    • Maintain separate analyses for urban and rural areas to identify geographical disparities

  5. Quality Control

    • Exclude districts with insufficient baseline data

    • Account for outliers in speed measurements

    • Consider only districts with consistent testing patterns

Limitations#

  • The data relies on user-generated tests

  • Test frequency can fluctuate over time

  • Geographic coverage depends on user participation

  • Test locations may not be consistently represented across time periods

Visualizations#

The analysis includes several visualizations:

  • Daily number of tests by region and district

  • Average download and upload speeds

  • Percentage changes in network performance

  • Impact analysis of the internet shutdown period (July 18-23, 2024)

Contents