pstatstools 0.0.10

Statistical tools package

pstatstools

A comprehensive Python package for statistical analysis, data visualization, and hypothesis testing with an intuitive API.

Overview

pstatstools is a user-friendly statistical toolkit that offers:

1. Simple but powerful Sample class for descriptive statistics, hypothesis testing, and visualization
2. Extensive probability distributions support through a flexible distribution API
3. Advanced categorical data analysis with contingency tables and chi-square tests
4. Non-parametric tests for when assumptions of parametric tests aren't met
5. Error propagation utilities for measurements with uncertainties
6. Beautiful visualizations that work seamlessly in Jupyter notebooks
7. ANOVA and inferential statistics for experimental design analysis

Quick Start

import pstatstools as pst
import matplotlib.pyplot as plt

# Create a sample with a single line of code
sample_data = [1, 4, 3, 5, 7, 3, 3, 2, 1, 5]
s = pst.sample(sample_data)

# Get beautiful HTML-formatted descriptive statistics in Jupyter notebooks
s.describe()  # Rich output with tables for central tendency, dispersion, etc.

# Visualize your data
s.histogram(bins=8, title="My Data Distribution")
plt.show()

# Run hypothesis tests
result = s.t_test(popmean=2.5)
print(f"T-test p-value: {result['p_value']:.4f}")

# Work with distributions
norm_dist = pst.distribution('normal', mu=0, sigma=1)
norm_dist.plot_pdf()
plt.show()

# Analyze categorical data
contingency_table = pst.categorical.ContingencyTable([[20, 15], [10, 25]])
chi2_result = contingency_table.chi_square_test()
print(f"Chi-square p-value: {chi2_result['p_value']:.4f}")

# Calculate with error propagation
result, uncertainty = pst.propagate_error_product(10.5, 0.2, 3.2, 0.1)
print(f"Result: {pst.format_with_uncertainty(result, uncertainty)}")

Installation

pip install pstatstools

The Amazing Sample Class

The Sample class is the heart of pstatstools - it wraps your data into an object with powerful statistical methods:

import pstatstools as pst
import numpy as np

# Create a sample from any iterable
data = [23, 25, 21, 24, 29, 22, 20, 25, 24, 27]
s = pst.sample(data)

# Basic statistics with intuitive methods
print(f"Mean: {s.mean():.2f}")         # 24.00
print(f"Median: {s.median():.2f}")     # 24.00
print(f"Std Dev: {s.std():.2f}")       # 2.62
print(f"95% CI: {s.ci()}")             # (22.17, 25.83)

# Beautiful formatted output in Jupyter notebooks
s.describe()  # Creates an elegant HTML table with all key statistics

Jupyter Notebook Integration

The describe() method generates rich, interactive output in Jupyter notebooks:

• Organized tables for central tendency, dispersion, quartiles, and shape statistics
• Automatic detection of notebook environment for optimal display
• Confidence intervals and interpretative guidelines
• Customizable precision with the round_to parameter

Visualization Made Simple

Creating statistical plots is effortless:

import matplotlib.pyplot as plt

# Histogram with normal curve overlay
s.histogram(bins=10, density=True, show_norm=True)

# Box plot
s.boxplot()

# Q-Q plot for normality check
s.qq_plot()

# Show all plots
plt.show()

Hypothesis Testing

Run common statistical tests with a single method call:

# One-sample t-test
result = s.t_test(popmean=22)
print(f"t-statistic: {result['t_statistic']:.2f}, p-value: {result['p_value']:.4f}")

# Compare with another sample
s2 = pst.sample([19, 18, 22, 20, 23, 21, 20, 19, 21, 20])
result = s.welch_test(s2)
print(f"Welch's t-test p-value: {result['p_value']:.4f}")

# Test for normality
normality = s.normality_test()
print(f"Is normal? {normality['is_normal']}")

Distribution Framework

Create and work with probability distributions using the intuitive distribution factory function:

# Create a normal distribution
norm = pst.distribution('normal', mu=100, sigma=15)

# Calculate probabilities
p = norm.cdf(115) - norm.cdf(85)  # P(85 < X < 115)
print(f"Probability between 85 and 115: {p:.4f}")

# Plot the distribution
norm.plot_pdf(shade_area=(85, 115))
plt.show()

# Compare with data
data = norm.random(1000)  # Generate 1000 random samples
norm.compare_with_data(data)
plt.show()

Categorical Data Analysis

import pstatstools as pst

# Create a contingency table
table = pst.categorical.ContingencyTable([
    [30, 10, 15],
    [25, 20, 10]
])

# Run chi-square test
result = table.chi_square_test()
print(f"Chi-square: {result['chi2_statistic']:.2f}, p-value: {result['p_value']:.4f}")

# Plot the table
table.plot(kind='heatmap')
plt.show()

Non-Parametric Tests

# Run Mann-Whitney U test (non-parametric t-test alternative)
result = pst.nonparametric.mann_whitney_test([1, 2, 3, 4, 5], [2, 3, 4, 5, 6])
print(f"Mann-Whitney U p-value: {result['p_value']:.4f}")

# Run Wilcoxon signed-rank test (non-parametric paired t-test alternative)
result = pst.nonparametric.wilcoxon_test([1, 2, 3, 4, 5], [1.1, 2.2, 3.3, 4.4, 5.5])
print(f"Wilcoxon p-value: {result['p_value']:.4f}")

Error Propagation

# Simple error propagation for product
result, uncertainty = pst.propagate_error_product(10.2, 0.1, 5.3, 0.2)
print(f"10.2 ± 0.1 × 5.3 ± 0.2 = {result:.2f} ± {uncertainty:.2f}")

# Monte Carlo error propagation for complex functions
def my_function(x, y, z):
    return (x**2 + y) / z

result, uncertainty = pst.monte_carlo_error_propagation(
    my_function,
    {'x': (5.0, 0.1), 'y': (10.0, 0.2), 'z': (2.0, 0.05)}
)
print(f"Result: {pst.format_with_uncertainty(result, uncertainty)}")

ANOVA and Experimental Design

# One-way ANOVA from summary data
result = pst.anova_from_summary_table(
    ss_total=500, 
    ss_between=200, 
    n_groups=3, 
    n_total=30
)
print(f"F-statistic: {result['f_statistic']:.2f}, p-value: {result['p_value']:.4f}")

# Power analysis for ANOVA
power = pst.anova_power_analysis(groups=3, n_per_group=10, effect_size=0.4)
print(f"Statistical power: {power:.3f}")

# Sample size calculation
sample_size = pst.calculate_sample_size_anova(groups=3, effect_size=0.3, power=0.8)
print(f"Required sample size: {sample_size}")

Probability Calculations

# Calculate binomial probability
p = pst.probability.binomial_probability(k=3, n=10, p=0.2)
print(f"P(X = 3) = {p:.4f}")

# Calculate poisson probability 
p = pst.probability.poisson_probability(k=5, lambda_=3, cumulative=True)
print(f"P(X ≤ 5) = {p:.4f}")

# Calculate waiting time in a Poisson process
stats = pst.probability.calculate_waiting_time(arrival_rate=2, n_arrivals=3)
print(f"Expected waiting time: {stats['mean']:.2f} units")
print(f"95% CI: ({stats['ci_95'][0]:.2f}, {stats['ci_95'][1]:.2f})")

Dependencies

• Python >= 3.6
• NumPy
• SciPy
• Matplotlib
• Pandas
• Statsmodels

Complete Documentation

For complete API documentation and more examples, visit our documentation site.

Contributing

Contributions are welcome! Please feel free to submit a pull request or open an issue.

License

Distributed under the terms of the MIT License.