Exploring Weather Data

import pandas as pd
import numpy as np

df = pd.DataFrame({
    "date":          ["2024-07-01","2024-07-02","2024-07-03","2024-07-04",
                      "2024-07-05","2024-07-06","2024-07-07","2024-07-08",
                      "2024-07-09","2024-07-10","2024-07-11","2024-07-12"],
    "temp_max_c":    [34.2, 36.8, 29.4, 27.1, 31.5, 38.4, 33.7, 35.1,
                      28.8, 32.6, 39.2, 30.4],
    "temp_min_c":    [21.8, 23.1, 18.6, 17.2, 20.4, 24.8, 22.1, 22.9,
                      18.1, 21.4, 25.6, 19.8],
    "humidity_pct":  [68, 74, 82, 88, 71, 78, 65, 72, 85, 69, 81, 76],
    "wind_kmh":      [14, 11, 28, 32, 18, 9, 22, 16, 35, 13, 8, 24],
    "rainfall_mm":   [0.0, 0.0, 12.4, 24.8, 2.1, 0.0, 0.0, 0.0,
                      18.6, 0.0, 0.0, 4.2],
    "solar_rad_wm2": [742, 698, 412, 298, 581, 814, 768, 731,
                      344, 695, 842, 624]
})

# Parse date
df["date"] = pd.to_datetime(df["date"])

print("Shape:", df.shape)
print("Missing values:", df.isnull().sum().sum())

# Temperature distribution
print("\nTemperature max (°C) — distribution:")
print(df["temp_max_c"].describe().round(1))

# Z-scores to identify extreme temperature days
mean_t = df["temp_max_c"].mean()
std_t  = df["temp_max_c"].std()
df["temp_zscore"] = ((df["temp_max_c"] - mean_t) / std_t).round(2)

print(f"\nMean max temp: {mean_t:.1f}°C | Std: {std_t:.1f}°C")
print("\nAll days sorted by temperature (hottest first):")
print(df[["date","temp_max_c","temp_zscore"]].sort_values(
    "temp_max_c", ascending=False
).to_string(index=False))

# Simplified Heat Index formula (Rothfusz approximation — Celsius version)
# Valid when temp >= 27°C and humidity >= 40%
# HI = -8.78469 + 1.61139411*T + 2.33854883*H - 0.14611605*T*H
#      - 0.01230809*T^2 - 0.01642482*H^2 + 0.00221173*T^2*H
#      + 0.00072546*T*H^2 - 0.00000358*T^2*H^2
# where T = temp_max_c, H = humidity_pct

T = df["temp_max_c"]
H = df["humidity_pct"]

df["heat_index_c"] = (
    -8.78469
    + 1.61139411  * T
    + 2.33854883  * H
    - 0.14611605  * T * H
    - 0.01230809  * T**2
    - 0.01642482  * H**2
    + 0.00221173  * T**2 * H
    + 0.00072546  * T * H**2
    - 0.00000358  * T**2 * H**2
).round(1)

# Safety flag: heat index >= 40°C triggers worker safety protocols
HEAT_SAFETY_THRESHOLD = 40.0
df["heat_alert"] = np.where(df["heat_index_c"] >= HEAT_SAFETY_THRESHOLD,
                             "ALERT", "OK")

print("Heat index and safety status:")
print(df[["date","temp_max_c","humidity_pct","heat_index_c",
          "heat_alert"]].sort_values("heat_index_c", ascending=False).to_string(index=False))

alert_days = df[df["heat_alert"] == "ALERT"]
print(f"\nDays triggering safety protocol (HI >= {HEAT_SAFETY_THRESHOLD}°C): {len(alert_days)}")
print(f"Max heat index recorded: {df['heat_index_c'].max():.1f}°C on {df.loc[df['heat_index_c'].idxmax(),'date'].date()}")

# Compute temperature range: max minus min for each day
df["temp_range_c"] = (df["temp_max_c"] - df["temp_min_c"]).round(1)

# Full correlation matrix across all numeric weather variables
numeric_cols = ["temp_max_c","temp_min_c","humidity_pct",
                "wind_kmh","rainfall_mm","solar_rad_wm2","temp_range_c"]

corr = df[numeric_cols].corr().round(3)
print("Correlation matrix:")
print(corr.to_string())

# Extract and rank key correlations with temp_max
print("\nKey correlations with temp_max_c:")
temp_corrs = corr["temp_max_c"].drop("temp_max_c").sort_values(key=abs, ascending=False)
for var, r in temp_corrs.items():
    print(f"  {var:<18} r = {r:+.3f}")

# Does wide daily temp range correlate with lower solar radiation?
range_solar = df["temp_range_c"].corr(df["solar_rad_wm2"]).round(3)
print(f"\nTemp range vs solar radiation: r = {range_solar:+.3f}")

# Define operational risk thresholds
HEAT_THRESH  = 36.0   # °C max temperature — elevated heat risk
WIND_THRESH  = 25.0   # km/h — risk for working at height
RAIN_THRESH  = 10.0   # mm — slippery surfaces, visibility
HI_THRESH    = 40.0   # °C heat index — safety protocol

# Create binary risk flags per condition
df["risk_heat"]  = (df["temp_max_c"]   >= HEAT_THRESH).astype(int)
df["risk_wind"]  = (df["wind_kmh"]     >= WIND_THRESH).astype(int)
df["risk_rain"]  = (df["rainfall_mm"]  >= RAIN_THRESH).astype(int)
df["risk_hi"]    = (df["heat_index_c"] >= HI_THRESH  ).astype(int)

# Composite risk score: sum of all active risk flags (0–4)
df["risk_score"] = df[["risk_heat","risk_wind","risk_rain","risk_hi"]].sum(axis=1)

# Risk level label
df["risk_level"] = pd.cut(
    df["risk_score"],
    bins=[-1, 0, 1, 2, 4],
    labels=["Low","Moderate","High","Critical"]
)

# Daily risk summary sorted by score
risk_summary = df[["date","temp_max_c","heat_index_c","wind_kmh",
                    "rainfall_mm","risk_score","risk_level"]].sort_values(
    "risk_score", ascending=False
)

print("Daily operational risk summary:")
print(risk_summary.to_string(index=False))

# Count by risk level
print("\nRisk level distribution:")
print(df["risk_level"].value_counts().sort_index().to_string())