A chart must make the answer to a task visible. Compute the answers first, so you can check the chart. Standard input: a CSV with columns date,<item>,… (the second column labels items, e.g. district), a line ---, then tasks TARGET ATTRIBUTE [ATTRIBUTE]:
extremes a→extremes pm25: min 35 (Ouhai, 2026-03-01), max 95 (Ouhai, 2026-03-03)(first occurrence);distribution a→distribution pm25: min 35, Q1 39.25, median 45, Q3 52, max 95(pandas quantiles,:g);outliers a→ values outside [Q1 − 1.5·IQR, Q3 + 1.5·IQR]:outliers pm25 (outside [20.125, 71.125]): Ouhai 03-03 (95)(ornone);trends a→ least-squares slope of the daily mean against day number:trends pm25: +5.75 per day (rising) over 4 days(rising if > 0.05, falling if < −0.05, else flat);correlation a b→correlation temperature pm25: r = -0.53 (moderate negative)(strong if |r| ≥ 0.7, moderate if ≥ 0.4, else weak).