# 👋 Welcome to my data story blog Hi, I'm **Reyna**. I use data to answer questions people actually ask: *Who will buy an electric vehicle? Does personality decide your income? Is your daily coffee really that expensive?* Every post follows the full path from **question → data → model → explanation → decision**. You'll see the charts, the methods, and the places where the data *can't* answer the question. ```mermaid flowchart LR Q[❓ Question] --> D[🗂️ Data & EDA] D --> M[🤖 Model] M --> E[🔍 Explain<br/>SHAP · effects] E --> A[✅ Decision] ``` --- ## 🧮 Try it yourself Interactive tools built from my models, running right in your browser: | Tool | What it does | |---|---| | [[Big Five Life Outcomes Calculator]] | Set a personality profile and see predicted income, job performance and life satisfaction | | [[Wealth Calculator]] | Explore how saving, investing and life factors relate to net worth | | [[🧮 Finance Calculator]] | Everyday personal-finance math | --- ## 📊 Case studies ### 🤖 Machine learning & prediction | Story | Question | Method | Headline | |---|---|---|---| | [[Who Will Buy an EV]] | Who's ready to switch to an electric car? | Gradient boosting + SHAP | Subsidy is the gate; the top 20% of scores hold 76% of buyers | | [[Big Five Personality & Life Outcomes]] | Does personality decide income, job performance and happiness? | Ridge, gradient boosting + SHAP | Personality drives 87% of life satisfaction but only 18% of income | | [[Is Wealth Earned or Inherited?]] | What builds wealth for the bottom 90% vs. the top 10%? | LightGBM + SHAP | Habits and homeownership for most; inheritance and investing at the top | | [[The Secret of Marriage Longevity]] | What predicts divorce? | Boosted trees + SHAP | Contempt and criticism outweigh demographics | | [[Predict Smartphone Addiction]] | Can phone habits predict addiction? | Logistic regression (691K users) | Social media time: 3.4× the odds per SD | | [[Four-legs Friends Forever Home]] | Which adopted dogs get returned, and why? | CatBoost + SHAP | Behavior beats breed; most returns happen within 2 weeks | | [[Diabetes Stage Prediction]] | Who progresses from pre-diabetes to Type 2? | Multiclass CatBoost | Family history, age and activity lead the risk | ### 🧩 Segmentation & behavior | Story | Question | Method | Headline | |---|---|---|---| | [[Credit Card Customers Segmentation]] | What kinds of card customers exist? | K-Means (50K customers) | Four personas, defined by behavior rather than demographics | | [[Understand Recovery]] | What really drives fitness recovery? | Regression + K-Means | HRV relative to *your own* baseline matters most | ### 📰 Data journalism & money math | Story | Question | Headline | |---|---|---| | [[Analytics Cases/Permits Are Not Homes\|Permits Are Not Homes]] | Will new housing permits bring rent relief within a year? | Permits are early signals; apartments often take 18+ months | | [[The Real Cost of Daily Coffee]] | Buy now or invest the money? | Price every purchase as an "experience premium" | --- ## 📚 Data science notebook The methods behind the stories, written as my own study notes. > [!example]- 🧪 Statistics & experimentation > - [[Statistical Knowledge]] · [[Bayesian Knowledge]] > - [[T Test]] · [[ANOVA Test]] > - [[Experiment Testing]] · [[Multivariate Experiment]] · [[CUPED]] > [!example]- 🔗 Causal inference > - [[Difference in Difference]] · [[Propensity Scoring Matching]] > - [[Uplift Modeling]] · [[Survival Analysis]] > [!example]- 🤖 Machine learning > - [[Regression]] · [[Logistic Regression]] · [[Tree-based Models]] · [[KNN]] > - [[KMeans]] · [[PCA]] > - [[Exploratory Data Analysis]] · [[Feature Selection]] · [[Numerical Features Data Process]] > - [[Machine Learning Knowledge]] · [[Machine Learning System Design]] > [!example]- 📈 Marketing & optimization > - [[Marketing Mixed Modeling]] · [[Product and Marketing Case]] > - [[Convex Optimization]] · [[Linear Programs]] --- > [!quote] How I read results > A model that explains 25% of the variance is still telling you something true: the other 75% is outside the data. I try to show both.