Hugging Face's logo

Datasets:
judicialmind
/
KCC-Pulse

Tasks:
Time Series Forecasting
Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
Size:
1M - 10M
Tags:
india
agriculture
kisan-call-centre
pests
crop-disease
surveillance
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
KCC-Pulse / README.md
sakasurya's picture
sakasurya
Access note: verified organisation email
bf4b954 verified
preview code
|
Raw History Blame Contribute Delete
9.17 kB
metadata 
license: cc-by-nc-sa-4.0
pretty_name: KCC-Pulse
language:
  - en
task_categories:
  - time-series-forecasting
tags:
  - india
  - agriculture
  - kisan-call-centre
  - pests
  - crop-disease
  - surveillance
  - aggregates
  - data.gov.in
size_categories:
  - 1M<n<10M
configs:
  - config_name: district_week_theme
    default: true
    data_files: panel/district_week_theme.parquet
  - config_name: state_week
    data_files: panel/state_week.parquet
  - config_name: series
    data_files: series/series.parquet
  - config_name: forecasts
    data_files: forecasts/forecasts.parquet
extra_gated_prompt: >-
  This dataset is released under the Creative Commons
  Attribution-NonCommercial-ShareAlike 4.0 licence (CC BY-NC-SA 4.0): research,
  academic, personal and other non-commercial use is permitted with attribution;
  adaptations must be shared under the same licence; commercial use is not
  permitted. The original public source data remains under the Government Open
  Data License - India (GODL-India). Commercial use: contact
  surya@judicialmind.ai or hello@judicialmind.ai.
extra_gated_fields:
  Name: text
  Email: text
  Organization: text
  I accept the CC BY-NC-SA licence terms (research and non-commercial use only): checkbox
extra_gated_button_content: Accept and request access

KCC-Pulse

Licence: CC BY-NC-SA 4.0 — research and non-commercial use only; derivatives must be shared under the same licence. Commercial use: contact surya@judicialmind.ai or hello@judicialmind.ai

Access is approved automatically when the email you enter matches your Hugging Face account email and belongs to the organisation you name; other requests are reviewed manually.

Weekly counts of Kisan Call Centre (KCC) calls per district and theme (pest, disease, weed, nutrient or intent), for the 10 best-covered states, December 2009 to July 2025, with 1-4 week forecast backtests and surge alerts. Aggregates only: no call text. A rise in calls about a pest or disease in a district is a signal of what farmers are dealing with that week.

Quickstart

The repository is gated: accept the terms on the dataset page, then huggingface-cli login.

from datasets import load_dataset

panel = load_dataset("judicialmind/KCC-Pulse", "district_week_theme", split="train").to_pandas()
pests = panel[(panel.theme_kind == "entity") & (panel.state == "MAHARASHTRA")]
print(pests.groupby("theme")["calls"].sum().sort_values(ascending=False).head(10))

With pandas or DuckDB on a local copy (huggingface-cli download judicialmind/KCC-Pulse --repo-type dataset --local-dir KCC-Pulse):

import duckdb, pandas as pd

state = pd.read_parquet("KCC-Pulse/panel/state_week.parquet")
print(state[state.state == "PUNJAB"].tail(3))

print(duckdb.sql("""
    SELECT state, district, week, theme, calls
    FROM 'KCC-Pulse/panel/district_week_theme.parquet'
    WHERE theme = 'pest_control' AND calls IS NOT NULL
    ORDER BY calls DESC LIMIT 5
""").df())

Series for forecasting are in series/series.parquet (series_id = STATE|DISTRICT|THEME, weekly ts, target); the evaluation origins and horizons are in splits.json.

Files

file rows contents
panel/district_week_theme.parquet 3,854,625 calls per district x ISO week (Monday) x theme
panel/state_week.parquet 7,920 all calls per state x week
series/series.parquet 4,629,053 5,959 weekly district x theme series used for forecasting, with the PM-KISAN covariate
forecasts/forecasts.parquet 200,000 backtest forecasts (quantiles 0.1-0.9 and mean) with the observed value: seasonal naive for all six origins and Poisson GLM for the origins February-June 2025
forecasts/summary.json forecast and alert scores per model (table below)
splits.json forecast cutoff, origins, horizons (1-4 weeks), season length (52)
leadlag.json cross-correlation between weekly calls about a crop and the weekly change in its mandi price, per district x crop

Schema

panel/district_week_theme.parquet

column type description
state, district string as in the KCC records (upper case); district = 0 when the call has no district
state_lgd, district_lgd int LGD codes where the district name matched
week date ISO week start (Monday)
theme string the pest, disease, weed or nutrient named in the calls (theme_kind = entity, 40,999 names), else the call intent (intent: weather, government_scheme, sowing_cultural, variety_seed, fertilizer_use, pest_control, nutrient_management, weed_management, market_price, disease_control, animal_husbandry, credit_insurance, irrigation_water, post_harvest, machinery, procurement_msp_complaint, other, unlabelled)
theme_kind string entity or intent
calls int number of calls; null when suppressed
calls_suppressed bool true when the count was 1-9

panel/state_week.parquet: state, week, state_calls, state_calls_suppressed.

series/series.parquet: series_id, ts (week), target (calls; null = no calls logged in the state that week, or suppressed), target_suppressed, state, district, district_lgd, theme, cov_pmkisan_beneficiaries (PM-KISAN beneficiaries in the district for the installment covering the week).

forecasts/forecasts.parquet: series_id, origin, h (horizon, weeks), model, ts, q0.1 ... q0.9, mean, y (observed calls), y_suppressed.

A week is missing (null), not zero, when the state logged no calls at all that week. Small cells: published call counts of 1-9 are null with <column>_suppressed = true (zeros are kept).

Results

Forecasts of weekly calls per district and theme, 5,959 series, 6 origins (2025-02-17 to 2025-07-07), horizons 1-4 weeks, 27,870 windows per model (forecasts/summary.json):

model WQL MASE alert precision alert recall
seasonal_naive 1.103 0.559 0.10 0.05
poisson_glm 0.917 0.676 0.73 0.39
chronos2 0.698 0.371 0.81 0.38

Alert = next-week calls more than 2x the seasonal baseline (same ISO week +-1 in the previous 3 years) and at least 10 calls. Forecast scores were computed on the unsuppressed counts.

Coverage

10 best-covered states: Rajasthan, Maharashtra, Gujarat, Bihar, Punjab, Odisha, Karnataka, Haryana, Madhya Pradesh, Himachal Pradesh; 356 district names (one per state is 0, calls without a district); weeks from 2009-12-28 to 2025-07-28. The source holds 27,754,152 calls from the public data.gov.in snapshot.

Source data

Derived from 2 resources published on the Open Government Data Platform India (data.gov.in), snapshot May 2026, under the Government Open Data License - India.

Providers: Ministry of Agriculture and Farmers Welfare.

Original data: published by Department of Agriculture and Farmers Welfare, Ministry of Agriculture and Farmers Welfare (Kisan Call Centre; PM-KISAN) on the Open Government Data Platform India (data.gov.in) and, where marked, NITI Aayog's National Data and Analytics Platform (NDAP), under the Government Open Data License - India (GODL-India). The original public data remains under GODL-India; the CC BY-NC-SA 4.0 licence of this repository covers JudicialMind's compilation, derived fields, annotations, benchmarks and models.

Citation 

@misc{judicialmind2026kccpulse,
  title={KCC-Pulse: weekly Kisan Call Centre call counts by district and theme, with forecasts and surge alerts},
  author={JudicialMind},
  year={2026},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/datasets/judicialmind/KCC-Pulse}}
}

License

CC BY-NC-SA 4.0 (see LICENSE): research, academic, personal and other non-commercial use, with attribution to JudicialMind and to the original data publishers; adaptations must be shared under the same licence. Commercial use requires a separate licence.

Commercial use: contact surya@judicialmind.ai or hello@judicialmind.ai.