A function that takes a pandas DataFrame as an input and returns a transposed DataFrame with the calculated mean, median, standard deviation, variation, skewness, coefficient of variation as percentage, mean/median difference, and kurtosis for each numeric column.
Exists as a single python file which can easily be imported as a funtion within your IDE.
Place super_describe.py in your project directory.
import pandas as pd
from super_describe import super_describe
# Load your data into a pandas DataFrame
df = pd.read_csv("path_to_your_file.csv")
# Generate statistical description
result = super_describe(df)
print(result)To enhance the table with additional columns for robustness and interpretation, here’s how you can modify it:
| Statistical Measure | Description | Units | Comparable across cols? | Robustness | More robust than |
|---|---|---|---|---|---|
| Mean | Average | Data | No | No - outliers | - |
| Mean Trimmed | Average after trimming extremes | Data | No | Yes - outliers | Mean |
| Mean Harmonic | Harmonic mean | Data | No | No - extreme values | - |
| Mean Geometric | Geometric mean | Data | No | Yes - extreme values | - |
| Median | Middle value | Data | No | Yes - outliers | Mean, Mean trimmed |
| Mode | Most frequent value | Data | No | Yes - outliers | Mean, Median |
| Standard Deviation (STDDEV) | Dispersion of values relative to mean | Data | No | No - outliers | MAD, IQR |
| STDDEV Relative % (RSD) | STDDEV as (% of mean) | % | Yes | - | - |
| Variance | STDDEV^2 | Data^2 | No | No - outliers | - |
| Variance Coefficient % | STDDEV / mean (as % or ratio) - Similar to RSD | % | Yes | - | - |
| Median Absolute Deviation (MAD) | Variability measure based on median (more robust) | Data | No | Yes - outliers | STDDEV, Variance |
| Standard Error of Mean (SEM) | STDDEV of sample-mean estimate | Data | No | - | - |
| Q1 (First Quartile) | Median of lower half | Data | No | Yes - outliers | - |
| Q3 (Third Quartile) | Median of upper half | Data | No | Yes - outliers | - |
| Skew | Asymmetry measure | Dimensionless | Yes | Partly - extreme values | - |
| Skew Coefficient | Degree of asymmetry of distribution around its mean | Dimensionless | Yes | - | - |
| Mean-Median Difference | Difference between mean and median | Data | No | No - outliers | - |
| Kurtosis | "Tailedness" measure | Dimensionless | Yes | Partly - extreme values | - |
| Kurtosis Excess | "Peakedness" compared to normal distribution | Dimensionless | Yes | - | - |
| Minimum | Smallest value | Data | No | Yes - outliers | - |
| Maximum | Largest value | Data | No | Yes - outliers | - |
| Outliers (Tukey Method) | Outliers count using Tukey's method (1.5 * IQR) | Count | No | Yes - skewed distributions | - |
| Outliers (1.96 Standard Deviation) | Outliers count using 1.96 * Std. Dev rule. | Count | No | Yes - symmetric distributions | - |