This Jupyter Notebook performs a stratification analysis on herbarium sample data. It evaluates whether the samples are correctly stratified into batches by comparing the distribution of taxonomic families and genera within each batch to the overall distribution in the original dataset. The analysis is performed using bar plots and Chi-square statistical tests. Contents
Data Loading and Preparation:
Load the expanded herbarium sample dataset.
Create a new column for sample counts.
Batch Generation:
Automatically calculate the number of batches based on the dataset size and a specified batch size.
Split the data into stratified batches.
Visualization:
Generate bar plots to visualize the distribution of families and genera in the overall dataset and in each batch.
Display the bar plots in a combined subplot layout for easy comparison.
Statistical Analysis:
Perform Chi-square tests to compare the observed and expected frequencies of families and genera in each batch.
Evaluate and interpret the results to determine if stratification was achieved.
Result Interpretation:
Print a summary message based on the Chi-square test p-values, indicating whether the stratification was successful.
How to Use
Clone the Repository:
sh
git clone https://github.com/yourusername/herbarium-stratification-analysis.git cd herbarium-stratification-analysis
Install Dependencies:
Ensure you have Jupyter Notebook installed. You can install it using pip:
sh
pip install jupyter
Install the required Python packages:
sh
pip install pandas plotly scipy
Run the Notebook:
Launch Jupyter Notebook:
sh
jupyter notebook
Open the Stratification_Analysis.ipynb notebook and run the cells.
Key Results
Bar Plots: Visual comparisons of the distribution of families and genera in the overall dataset and in each batch.
Chi-square Test Results: Statistical evaluation showing p-values and Chi-square statistics for each batch.
Interpretation Message: A summary message that indicates whether the observed distributions in the batches match the expected distributions derived from the overall dataset.
Example Chi-square Test Results batch chi2_family p_family chi2_genus p_genus 1 0.19 1.00 0.28 1.00 2 0.32 1.00 0.58 1.00 3 0.07 1.00 0.17 1.00
High p-values (close to 1) suggest that there is no significant difference between the observed and expected distributions for families and genera in each batch. This means that the observed frequencies in your batches are very close to the expected frequencies derived from the overall dataset. License
This project is licensed under the MIT License. See the LICENSE file for details.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent