data trove noaa lightning strikes 2018

This column is a geographic latitude, with values ranging from 25.35 to 35.46 degrees north — consistent with a mid-latitude region such as the Arabian Peninsula, South Asia, or the southern US Sun Belt. With only 868 unique values across 59,070 rows, latitudes are heavily discretised (likely snapped to a grid or centroid), not continuous GPS readings. The distribution is right-skewed (skew 1.15) with 9.5% of rows flagged as outliers (5,626 records), suggesting a dominant cluster around 27–30° with a long upper tail toward 35.46° that warrants geographic investigation. Treatment: Pair with longitude for spatial joins or clustering; investigate the 5,626 outlier records above ~32° for data-quality or sub-population issues before modelling. high · anthropic:default

n

59,070

nulls

0 (0.0%)

unique

868

min

25.35

max

35.46

mean

29.23

median

28.84

std

1.899

q1

27.84

q3

29.94

iqr

2.1

skew

1.147

kurtosis

0.7033

n_outliers

5,626

outlier_rate

0.09524

zero_rate

0

This column contains longitude coordinates, with all values falling between -96.74 and -79.03, consistent with the central/eastern United States (roughly spanning from Texas/Oklahoma eastward to the Great Lakes or Southeast). The distribution is moderately left-skewed (skew = -0.85) with mass concentrated toward the eastern end of the range (median -82.06, Q3 -81.22), suggesting most records originate from states like Ohio, Michigan, Florida, or nearby. Only 1,231 unique values across 59,070 rows indicates coordinates are discretized or snapped to a coarse grid rather than true GPS precision — an analyst should be aware this limits spatial resolution. Treatment: Use as-is for spatial joins or clustering; note limited precision (1,231 unique values for 59,070 rows) before any fine-grained geospatial analysis. high · anthropic:default

n

59,070

nulls

0 (0.0%)

unique

1,231

min

-96.74

max

-79.03

mean

-85.27

median

-82.06

std

5.252

q1

-89.94

q3

-81.22

iqr

8.72

skew

-0.8513

kurtosis

-0.8532

n_outliers

0

outlier_rate

0

zero_rate

0

This column encodes calendar month as an integer from 1 to 12, with exactly 12 unique values and no nulls across 59,070 rows. Distribution is nearly uniform (mean 6.90, median 7.0, skew −0.15, kurtosis −0.21), suggesting fairly even coverage across all months with a slight lean toward the second half of the year. No outliers or zero values are present, and the near-flat kurtosis rules out strong seasonal concentration. The even spread is mildly surprising if the underlying data were expected to show seasonal patterns. Treatment: Treat as a cyclic ordinal feature; apply sine/cosine encoding (sin(2π·month/12), cos(2π·month/12)) before modelling to capture cyclical continuity. high · anthropic:default

n

59,070

nulls

0 (0.0%)

unique

12

min

1

max

12

mean

6.899

median

7

std

2.335

q1

5

q3

8

iqr

3

skew

-0.155

kurtosis

-0.2056

n_outliers

0

outlier_rate

0

zero_rate

0

This column almost certainly represents a count of strikes (e.g., bowling, lightning, labor, or similar discrete events) per record, with values ranging from 1 to 531 across 381 unique integers. The distribution is heavily right-skewed (skew = 2.02, kurtosis = 6.09): the median is only 34 while the mean is 49.43 and the std is 49.83, indicating a long upper tail. Notably, 4.9% of rows (2,917) are flagged as outliers, and the IQR of 54 spans Q1=14 to Q3=68, confirming most records cluster low while a minority have very high counts up to 531. Treatment: Log-transform (log1p) before regression or distance-based modelling to reduce skew; investigate outlier group (n=2,917) for data quality or domain-specific segmentation. medium · anthropic:default

n

59,070

nulls

0 (0.0%)

unique

381

min

1

max

531

mean

49.43

median

34

std

49.83

q1

14

q3

68

iqr

54

skew

2.023

kurtosis

6.094

n_outliers

2,917

outlier_rate

0.04938

zero_rate

0