README.md

metadata

license: cc-by-4.0
language:
  - en
tags:
  - india
  - telecom
  - trai
  - wireless
  - broadband
  - mnp
  - subscribers
  - time-series
pretty_name: India Telecom Subscription Data (TRAI) 2016–2026
size_categories:
  - n<1K

India Telecom Subscription Data (TRAI) 2016–2026

Monthly telecom subscription statistics for India, parsed from the official Telecom Regulatory Authority of India (TRAI) Telecom Subscription Data (TSD) PDF reports.

Covers the full post-Jio era: from Jio's disruptive launch in September 2016 through to January 2026, capturing the collapse of smaller operators, the Vodafone-Idea merger, and the rise of wireless broadband.

Files

File	Rows	Description
data/telecom_monthly.parquet	119	National-level metrics, one row per month
data/operators_monthly.parquet	1,679	Wireless subscribers by operator × LSA × month (2025 onwards)

Loading the Data

import pandas as pd

monthly = pd.read_parquet("data/telecom_monthly.parquet")
operators = pd.read_parquet("data/operators_monthly.parquet")

telecom_monthly — Field Reference

Field	Unit	Coverage	Description
data_month	YYYY-MM	119/119	Reporting month
wireless_total_mn	millions	119/119	Total wireless (mobile) subscribers
wireline_total_mn	millions	119/119	Total wireline (fixed-line) subscribers
total_subscribers_mn	millions	119/119	Wireless + wireline combined
broadband_total_mn	millions	119/119	Total broadband subscribers
broadband_wireless_mn	millions	112/119	Wireless broadband subscribers
broadband_wireline_mn	millions	112/119	Wireline broadband subscribers
urban_wireless_mn	millions	112/119	Urban wireless subscribers
rural_wireless_mn	millions	112/119	Rural wireless subscribers
wireless_growth_pct	%	77/119	Monthly wireless subscriber growth rate
overall_tele_density_pct	%	67/119	Overall tele-density (subscribers per 100 population)
m2m_total_mn	millions	19/119	Machine-to-Machine (IoT) connections — available from mid-2024
mnp_monthly_mn	millions	92/119	Mobile Number Portability requests in the month
validation_score	0–1	119/119	Automated data quality score (see Methodology)
validation_status	pass/warn	119/119	Quality status

operators_monthly — Field Reference

Field	Unit	Description
data_month	YYYY-MM	Reporting month
lsa	string	Licensed Service Area (telecom circle) — 22 values
operator	string	Operator name (Airtel, Jio, Vi, BSNL, MTNL, Reliance Com.)
subscribers	millions	Wireless subscribers for this operator in this LSA
prev_month	millions	Previous month's subscriber count (for MoM comparison)
net_add	millions	Net subscriber addition / loss

LSAs (Licensed Service Areas)

Andhra Pradesh, Assam, Bihar, Delhi, Gujarat, Haryana, Himachal Pradesh, Jammu & Kashmir, Karnataka, Kerala, Kolkata, Madhya Pradesh, Maharashtra, Mumbai, North East, Odisha, Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh (E), Uttar Pradesh (W), West Bengal

Coverage Notes

What is complete (2016–2026):

• wireless_total_mn, wireline_total_mn, broadband_total_mn, total_subscribers_mn — fully populated for all 119 months

Structural gaps (TRAI reporting limitations, not parsing errors):

• wireless_growth_pct, overall_tele_density_pct — absent from 2016–2017 PDFs (different format era)
• m2m_total_mn — TRAI only began publishing M2M data prominently from mid-2024
• mnp_monthly_mn — absent for 2019–2020 (section format changed; data not extractable)
• mnp_zone1_mn, mnp_zone2_mn — not yet implemented
• Operator-level LSA data — TRAI only introduced the detailed wireless subscriber annexure (Annexure-II) from January 2025

Missing months:

• 2020-08 and 2021-12: not published by TRAI

Methodology

Data is extracted from TRAI's official PDF reports using a custom Python pipeline:

1. fetch_trai_index.py — scrapes TRAI's website for all report URLs
2. download_trai_pdfs.py — downloads and caches PDFs
3. parse_trai_pdf.py — extracts tables using pdfplumber; falls back to OCR (pytesseract) for scanned PDFs
4. validate_trai_month.py — runs 5 automated checks per month:
   • A (30%) Internal arithmetic (operator sums, broadband components)
   • B (30%) Cross-text plausibility (table vs. narrative text in same PDF)
   • C (20%) Range bounds (known historical ranges by era)
   • D (10%) Month-over-month continuity
   • E (10%) Structural completeness (22 LSAs, required annexures present)
5. build_trai_dataset.py — assembles parquets; applies manual patches for months with unextractable tables

All 119 months have validation score ≥ 0.79 (mean: 0.97).

Source

TRAI Telecom Subscription Data reports: https://www.trai.gov.in/release-publication/reports/telecom-subscriptions-reports

Data is derived from publicly available government documents. Original reports are the copyright of TRAI / Government of India.

License

This dataset is released under Creative Commons Attribution 4.0 International (CC BY 4.0).

You are free to share and adapt the data for any purpose, including commercial use, provided you give appropriate credit.

Suggested citation:

TRAI India Telecom Subscription Dataset (2016–2026).
Compiled from TRAI Telecom Subscription Data PDF reports.
https://huggingface.co/datasets/rahulmatthan/india-telecom-data
CC BY 4.0

Known Issues

• 2025-09: Airtel and BSNL subscriber counts are null for several LSAs due to a font-encoding issue in the source PDF
• 2025-10: 96 of 132 expected operator×LSA rows present (text-fallback parsing limitation)
• 2017-04: validation_score = 0.79 (lowest in dataset) — broadband component sum mismatch in source PDF