A Data-Driven Forecasting and Capacity-Prioritization Framework for Urban Energy Infrastructure: A Case Study of Almaty — Code and Data

This repository contains the full reproducible pipeline, data, and results accompanying the manuscript "A Data-Driven Forecasting and Capacity-Prioritization Framework for Urban Energy Infrastructure: A Case Study of Almaty" (submitted to MDPI Technologies). The pipeline forecasts residential energy demand growth and supports substation capacity planning for Almaty, Kazakhstan, using only publicly available real-estate marketplace data, official planning documents, and open-source tools. Contents: - scripts/ — the full pipeline (step1–step10): data collection, parsing, geocoding, Poisson/negative-binomial/random-forest/ensemble forecasting, out-of-sample backtesting, MILP-based substation-capacity optimization with Monte Carlo sensitivity analysis, and cross-validation against the confirmed construction pipeline. - data/ — residential-complex records scraped from krisha.kz (627 city-wide complexes, 128 in the three fast-growing districts of Alatau, Naurizbay, and Turksib), collected September 2026 in accordance with the site's terms of use. No personal data is included. - results/ — forecast outputs, backtest results, optimization results, and sensitivity-analysis results as reported in the manuscript. District selection was triangulated across three independent signals (marketplace growth data, the city administration's 2026 network-modernization plan, and the confirmed under-construction project pipeline). Five forecasting models were compared via out-of-sample backtesting; the resulting demand forecast feeds a mixed-integer linear program for substation-capacity allocation, validated with a Monte Carlo sensitivity analysis and cross-checked against official city-wide statistics from the Bureau of National Statistics of the Republic of Kazakhstan. See README.md for the full pipeline description and instructions to reproduce all results.