How Janadesh Pulse Works
The full methodology behind India Political Pulse — what feeds the score, how it's weighted, and what it deliberately does not claim to be.
What Janadesh Pulse Is
Janadesh Pulse is Janadesh Online’s own daily analysis of political momentum between the NDA and INDIA blocs, built entirely from free, public data. It is not a scientific opinion poll, not a seat projection, and not a guaranteed election prediction. Janadesh Pulse does not use published third-party opinion polls as a scoring input — every number behind it comes from Janadesh’s own election database, free news-monitoring APIs, or official YouTube activity.
Signal Categories & Weights
Fixed weights, read live from the scoring engine — not duplicated as hardcoded copy
Recent Election Performance
40%Vote-share change between the two most recent comparable elections (same house, same state or national scope) for each bloc's contesting parties — read from Janadesh's own verified election database, never aggregated across unrelated state assemblies.
News Attention
30%Each bloc's share of English-language news coverage, combining Google News RSS headline counts and GDELT's free news-monitoring API volume.
Leader Momentum
20%Tracked national leaders' own news-attention share, rolled up to bloc level through each leader's real, current alliance membership — never a PM-preference poll or an approval rating.
Digital Momentum
10%Upload activity share across official YouTube channels for each bloc's most-searched terms, via the YouTube Data API.
If a category has no usable data on a given day, its weight is redistributed proportionally across the categories that do have data — a missing category is never silently treated as a zero score.
Update Cadence
Data collection, scoring and summary generation run once a day as a single automated pipeline (recommended around 02:00 IST, before the day’s news cycle picks up). A reading is only used in that day’s score if it is less than 72 hours old — a source that has gone quiet drops out of the score rather than serving stale data as if it were current.
Confidence Logic
Confidence reflects how many of the four scored categories had usable, fresh data on a given day — high confidence means all four contributed; lower confidence means the score leaned more heavily on fewer signals. The NDA/INDIA classification itself uses two different gap thresholds (one to newly declare an advantage, a smaller one to give it up) so a single noisy day cannot flip the headline reading back and forth.
Why News Attention Isn't Voter Support
A party or leader dominating headlines can mean genuine momentum — or a scandal, a controversy, or an unrelated news cycle. News Attention and Leader Momentum measure how much a bloc or leader is being talked about, not whether that coverage is favorable, and never how people would actually vote.
Why Digital Activity Isn't Voting Intention
Digital Momentum counts upload activity on official YouTube channels — a proxy for campaign communication effort, not for how many people watched, agreed with, or were persuaded by that content. A bloc can post more videos without gaining a single vote.
Why the Reader Poll Is Excluded from Scoring
The Janadesh Reader Poll and Issue Poll are unscientific, self-selecting site-visitor votes — anyone can vote once a day, with no sampling, weighting, or demographic correction. They are kept in a separate database table from the real Pulse score and can never be joined into it, even accidentally. They exist purely as an engagement feature, not as evidence of anything beyond this site’s own visitors’ stated preference.
Why Trending Issues Don't Feed the Score
Trending Issues shows real headline-keyword matches, week-over-week change, and which tracked leaders are most co-mentioned with each issue — but there is no honest way to say an issue “belongs” to one bloc over the other, so it stays purely informational.
Known Limitations
GDELT’s news-volume signal is restricted to English-language coverage for consistency. This means Pulse systematically under-represents Hindi and other regional-language news cycles, which can dominate coverage of state-specific leaders and issues.
An earlier version of this page discussed a “state concentration” signal for Trending Issues. Headline data carries no state tagging today, and inventing one would mean guessing — so that signal has been dropped rather than faked, and issues are shown nationally only.
A topic needs at least 3 matching headlines in a 7-day window before Trending Issues will surface it — a 1-2 headline match is treated as noise, not a real issue signal.
Why Janadesh Pulse Is Not an Opinion Poll
Janadesh Pulse does not survey anyone, does not sample the electorate, and does not project seats or vote share. It is a rule-based reading of public attention and recent election performance — useful as one input among many, but not a forecast, a prediction, or a substitute for a scientific opinion poll.
