Computational Analysis of Autistic Women's Survey Responses: Data Preprocessing, Logistic Regression, Natural Language Processing, and Visualization

This repository contains the computational resources developed for the analysis of survey responses from autistic women, focusing on demographic characteristics, self-reported experiences, and open-ended narratives. The repository provides the datasets, Jupyter notebooks, and Python scripts used to perform data preprocessing, exploratory analysis, logistic regression, natural language processing (NLP), and survey response visualization. The research workflow is organized into five components: 00. Dataset Contains the original survey responses and a filtered dataset comprising demographic information and open-ended comments from autistic women. These datasets serve as inputs for the subsequent analytical procedures. 01. Preprocessing and ETL Contains Jupyter notebooks for data cleaning, transformation, and exploratory descriptive analysis. The preprocessing procedures prepare the survey responses for subsequent statistical and computational analyses. 02. Logistic Regression Contains the Jupyter notebook implementing logistic regression models for classification and evaluation based on the processed survey responses. The analysis includes data preparation, feature standardization, model training, feature selection, and performance evaluation. 03. Natural Language Processing (NLP) Contains the Jupyter notebook dedicated to analyzing open-ended survey responses. The workflow includes text preprocessing, semantic representation using sentence embeddings, clustering, dimensionality reduction, and the identification of representative linguistic patterns. 04. Survey Response Visualization Contains the Python script used to generate visual representations of survey response distributions across questionnaire categories, including an integrated 5 × 2 figure. Purpose and Scope The repository aims to provide transparency regarding the computational methods applied to the survey data and facilitate the inspection and reproduction of the analytical procedures. The materials are intended for research and methodological purposes. The analyses provide complementary perspectives on demographic characteristics, quantitative survey responses, and qualitative textual information. Software and Methods Python Jupyter Notebook pandas and NumPy scikit-learn spaCy Sentence Transformers (all-MiniLM-L6-v2) PCA and UMAP K-Means clustering Matplotlib and Seaborn Reproducibility The notebooks and scripts document the main stages of the computational workflow. Execution requires the corresponding input datasets and the Python libraries used in each analysis. File paths may need to be adjusted according to the local repository location.