JLM Air Quality Analysis

JLM-Air-Quality-Analysis is a Python repository that provides the analysis and visualisation layer for the Jerusalem-Air-Quality-Shabbat Hugging Face dataset. It tests the hypothesis that the reduction in traffic during Shabbat and Yom Tov in Jerusalem produces a measurable, predictable dip in traffic-derived pollutants such as NO, NO2, and NOx. The project replaces the dataset's calendar-based weekend flags with exact windows derived from Hebcal candle-lighting and havdalah times, classifies readings into mutually exclusive segments such as weekday, pre-Shabbat, Shabbat, post-havdalah, and Yom Kippur, and reports a headline comparison showing an approximately 62 percent median NO2 drop on Shabbat relative to the workweek, compared with 31 percent in New York and 20 percent in London over the same twelve-month window.

The pipeline is built with pandas, pyarrow, numpy, and matplotlib, and is organised into three scripts: one for halachic relabeling, one for building tidy control-city data from OpenAQ exports, and one for generating nine figures and a cross-city CSV. The repository documents its caveats directly, including confounding from meteorology, seasonality, and Saharan dust events, as well as the heuristic nature of the two-hour pre-Shabbat and post-havdalah windows.

The project may be of interest to researchers studying urban emissions and natural experiments, data analysts working with time-series environmental data, and developers interested in applying calendar-aware labeling to public datasets. It also serves as an example of reproducible analysis with explicit documentation of data sources, assumptions, and limitations.