Gujarat Vidhan Sabha 50 Granular Clusters & Political Wave Theory

1. 50 Granular Micro/Meso Clusters

Rather than broad macro-regions, this model groups Gujarat's 182 constituencies into 50 micro-regional clusters (averaging ~3 to 5 seats each). This captures micro-caste demographics (e.g. Saurashtra Patidar vs. Charotar Patidar vs. coastal Koli), industrial agglomeration (Morbi ceramics, Surat diamond/textiles, Bharuch chemical SEZ), and tribal forest belts (Aravalli, Panchmahal, Dangs).

2. 2008 Delimitation Reconciled (2002 Excluded)

In accordance with electoral data science standards:

• 2002 Election is 100% excluded from correlation, swing velocity, and wave vector computations.
• For the 132 continuous seats, 2007 election data is directly incorporated.
• For the 50 delimited seats created in 2008 (e.g. Ghatlodia, Vejalpur, Nikol, Bapunagar, Anklav, Tharad, etc.), we mapped their 2007 record to their official predecessor parent seat (e.g. Sarkhej, Naroda, Shaherkotda) with an explicit predecessor tag.
• The primary swing vectors are computed on the 3 identical post-delimitation elections: 2012, 2017, and 2022.

3. Electoral Swing Velocity & Wave Direction

For each seat \( i \), we compute the 2-dimensional swing vector:

Seats with earlier swing inflection and higher momentum act as radiators, directing waves toward downstream neighbor seats within radius \( d \le 2 \) crosses.

4. Strict Out-Degree Constraint (out_degree ≤ 1)

Every single constituency emits at most one outgoing wave (\( \text{out\_degree} \le 1 \)). Multiple incoming waves (\( \text{in\_degree} \ge 0 \)) or terminal sinks (\( \text{out\_degree} = 0 \)) are permitted, enforcing a directed forest rooted at cluster epicenters.