Special programming languages

Topic: "Visualization of statistical tests" Bulatova Viktoriia, Popp Sofia, SA-32

RR_PROJECT_BULATOVA_POPP

A Python statistical test visualization project designed to help analyze and visualize the results of statistical methods such as t-tests, ANOVA, and others.

Description

RR_PROJECT_BULATOVA_POPP is a comprehensive Python library designed for automated statistical hypothesis testing and data visualization. The package simplifies the decision-making process by automatically selecting the appropriate parametric or non-parametric test based on the data characteristics, including normality and sample size. It also provides robust visualizations to ensure clear and insightful presentation of results.

Main features:

1. Automatic Test Selection

• Automatically chooses between parametric and non-parametric tests based on:
  • Normality of data, determined using Shapiro-Wilk or D’Agostino-Pearson tests depending on the sample size.
  • Homogeneity of variances for multi-group tests.
  • Supports a wide range of statistical tests:
  • Paired samples: Paired t-test, Wilcoxon signed-rank test.
  • Independent samples: Independent t-test, Mann-Whitney U test.
  • Multi-group comparisons: One-way ANOVA, Alexander-Govern test, Kruskal-Wallis test.

2. Assumption Checks

• Normality Testing:
  • Shapiro-Wilk test for small samples.
  • D’Agostino-Pearson test for larger datasets.
  • Homoscedasticity (Equal Variances):
  • Bartlett’s test for verifying variance equality across groups.

3. Descriptive Statistics

• Provides detailed summaries for each sample:
  • Mean, median, mode, variance, standard deviation.
  • Skewness, kurtosis, range, minimum, and maximum.

4. Advanced Visualizations

• Visualizations are integrated into the analysis to ensure intuitive understanding:
  • Histograms with KDE: Displays data distributions and overlays the ideal normal curve.
  • Q-Q Plots: For assessing normality.
  • Boxplots: Highlights variability and spread across groups.
  • Joyplots: Overlays group distributions, emphasizing mean or median values for quick interpretation.
  • Custom Visualization for Differences: Specialized plots for paired tests, showing distributions and differences between samples.

Installation

To install this package, use pip install rr_project_bulatova_popp

Examples

Example 1

from rr_project_bulatova_popp import PairedTtest

df = pd.read_csv('..\\Datasets\\dataset_1.csv') 

paired_test = PairedTtest(df, ['col1', 'col2'])

paired_test.normality_visualization()

Result:

Example 2

from rr_project_bulatova_popp import PairedTtest

df = pd.read_csv('..\\Datasets\\dataset_1.csv') 

multi_group1 = MultiGroupTest(df, ['sem2_alc3', 'sem2_alc2', 'sem2_alc5'])

multi_group1.run_test()

Result:

---The Shapiro-Wilk normality test has been performed---

          sem2_alc3 sem2_alc2 sem2_alc5
stat       0.958085  0.954866  0.940237
p_value     0.31377  0.261891  0.112078
is_normal      True      True      True
All THE SAMPLES ARE NORMALLY DISTRIBUTED

---The one-way ANOVA test has been performed---

H0: Two or more groups have the same population mean