Divorce Prediction Analysis: A Study on Gottman's Couples Therapy Data

This project analyzes a dataset of 170 Turkish couples to identify the key statistical predictors of divorce. The analysis is based on the "Divorce Predictors Scale" (DPS) derived from Gottman's couples therapy principles.

• See the full technical analysis: divorce_analysis.ipynb
• See the summary presentation: presentation/presentation.pdf

Project Context

The dataset, sourced from Kaggle, contains responses to 54 questions from couples who were either "happily married" (Class 0) or "divorced" (Class 1). All responses were collected on a 5-point scale, where a higher score indicates a higher prevalence of problems.

The goal of this analysis was to move beyond individual questions and identify the themes that were most strongly correlated with divorce.

Data Source: Gottman Divorce Predictors Dataset on Kaggle

---

Key Findings

The analysis revealed a clear distinction between two types of marital problems: "Foundation Failures" and "Communication Breakdowns."

While both are factors in divorce, their predictive power is dramatically different.

1. The Strongest Predictors: "Foundation Failures" (Corr: 0.89 - 0.94)

These 10 questions represent a fundamental misalignment in the relationship's core. They are the strongest statistical predictors of divorce. A high score on these indicates a lack of shared values, goals, and emotional harmony.

/presentation/graphs/most_influent_questions.png

2. The Weakest Predictors: "Communication Breakdowns" (Corr: 0.42 - 0.71)

This group of 10 questions perfectly describes the destructive communication patterns (Contempt, Defensiveness, Stonewalling) identified by Gottman.

/presentation/graphs/least_influent_questions.png

The Big Idea:

While poor communication (like "The Four Horsemen") clearly erodes a marriage, a lack of a shared foundation (values, dreams, roles) is the strongest predictor of divorce.

This suggests that while communication can be repaired, a fundamental misalignment of values makes a stable marriage difficult to build from the start.

---

Tech Stack

• Python 3
• Pandas: For data loading and manipulation.
• Seaborn & Matplotlib: For data visualization.
• Jupyter Notebook: For the analysis environment.

---

How to Run This Project

1. Clone this repository:

   git clone [https://github.com/peptesta/divorce_analysis.git](https://github.com/peptesta/divorce_analysis.git)

2. If you want you can use the already downloaded dataset, or you can download it from the Kaggle link.
3. Place the divorce_data.csv and reference.tsv files into a /data folder (you may need to create this folder).
4. Create the /graphs folder (in order to get real time graphs about the data)
5. Open divorce_analysis.ipynb in a Jupyter environment to view and run the full analysis.