R Econometrics MCP Server

# Advanced Claude Desktop Examples with RMCP This guide provides concrete examples you can type directly into Claude Desktop to showcase RMCP's advanced statistical capabilities enabled by the comprehensive 429-package ecosystem. ## Prerequisites Make sure RMCP is configured in your Claude Desktop: ```json { "mcpServers": { "rmcp": { "command": "rmcp", "args": ["start"] } } } ``` ## 🧠 Machine Learning Examples ### K-means Customer Segmentation ``` I have customer data with variables: annual_spending, visit_frequency, avg_order_value, customer_age. Can you perform k-means clustering to segment customers into 3 groups and tell me what each segment represents? Use this sample data: annual_spending,visit_frequency,avg_order_value,customer_age 2400,24,100,28 4800,48,200,45 1200,12,50,22 3600,36,150,38 6000,60,300,52 1800,18,75,25 5400,54,250,48 2100,21,85,32 3900,39,175,41 4500,45,225,44 ``` **Expected Output**: Cluster assignments, centroids for each segment (e.g., "Budget Shoppers", "Premium Customers", "Occasional Buyers"), and business insights for targeting each segment. ### Random Forest for House Price Prediction ``` I want to predict house prices using a random forest model. I have these variables: bedrooms, bathrooms, sqft, age, garage_spaces, lot_size. Build a model and tell me which features are most important for pricing. Sample data: bedrooms,bathrooms,sqft,age,garage_spaces,lot_size,price 3,2,1800,15,2,0.25,320000 4,3,2400,8,2,0.35,450000 2,1,1200,25,1,0.15,250000 5,4,3200,5,3,0.5,650000 3,2.5,2000,12,2,0.3,380000 4,2,2200,20,2,0.4,420000 6,5,4000,3,3,0.75,850000 ``` **Expected Output**: Variable importance rankings, out-of-bag error rate, prediction accuracy, and practical insights about what drives home values. ### Decision Tree for Loan Approval ``` Create an interpretable decision tree model to predict loan approval (approved/denied) based on: credit_score, income, debt_ratio, employment_years, loan_amount. I want to understand the decision rules. Data: credit_score,income,debt_ratio,employment_years,loan_amount,approved 720,65000,0.3,5,200000,approved 580,35000,0.8,2,150000,denied 750,85000,0.2,8,300000,approved 620,45000,0.6,3,180000,denied 680,55000,0.4,6,220000,approved 540,25000,0.9,1,100000,denied 780,95000,0.1,10,350000,approved ``` **Expected Output**: Decision tree rules in plain English (e.g., "If credit_score > 650 AND debt_ratio < 0.5, then approve"), accuracy metrics, and variable importance. ## 📈 Advanced Econometrics Examples ### Panel Data Regression ``` I have panel data for 5 states over 4 years studying the effect of minimum wage on employment. Run a fixed effects panel regression to estimate the causal impact. Variables: state_id, year, min_wage, employment_rate, gdp_growth, population Data: state_id,year,min_wage,employment_rate,gdp_growth,population 1,2020,7.25,92.5,2.1,5000000 1,2021,8.00,91.8,1.8,5050000 1,2022,8.50,91.2,2.4,5100000 1,2023,9.00,90.9,3.1,5150000 2,2020,8.50,93.2,2.8,3000000 2,2021,9.25,92.7,2.2,3030000 2,2022,10.00,92.1,2.9,3060000 2,2023,10.75,91.6,3.4,3090000 ``` **Expected Output**: Fixed effects coefficients, statistical significance, interpretation of minimum wage impact on employment, R-squared values. ### Vector Autoregression (VAR) Model ``` Build a VAR model to analyze the dynamic relationships between GDP growth, inflation, and unemployment using quarterly data. I want to understand how shocks in one variable affect the others. Data: quarter,gdp_growth,inflation,unemployment 2020Q1,2.1,1.8,3.5 2020Q2,-9.5,0.1,14.7 2020Q3,33.4,1.2,8.4 2020Q4,4.3,1.4,6.7 2021Q1,6.3,2.6,6.0 2021Q2,6.7,5.4,5.9 2021Q3,2.3,5.3,4.8 2021Q4,6.9,6.8,3.9 2022Q1,-1.6,8.5,3.6 2022Q2,-0.6,9.1,3.6 ``` **Expected Output**: VAR coefficients for each equation, impulse response functions, variance decomposition, and economic interpretation of dynamic relationships. ### Instrumental Variables Regression ``` Estimate the causal effect of education on wages using distance to college as an instrument. I suspect education is endogenous due to ability bias. Formula: "wage ~ education + experience + age | distance_to_college + experience + age" Data: wage,education,experience,age,distance_to_college 45000,12,5,27,25 65000,16,8,32,5 35000,10,3,23,45 85000,18,12,38,2 50000,14,6,30,15 75000,16,10,34,8 40000,11,4,25,35 95000,20,15,43,3 ``` **Expected Output**: 2SLS estimates, first-stage statistics, weak instruments test, Wu-Hausman endogeneity test, and interpretation of causal education premium. ## ⏰ Time Series Forecasting Examples ### ARIMA Forecasting ``` I have monthly sales data and need to forecast the next 6 months. Build an ARIMA model and provide confidence intervals for the predictions. Data (monthly sales in thousands): month,sales 2023-01,145 2023-02,158 2023-03,162 2023-04,148 2023-05,171 2023-06,185 2023-07,178 2023-08,192 2023-09,186 2023-10,201 2023-11,198 2023-12,215 2024-01,156 2024-02,169 2024-03,174 ``` **Expected Output**: ARIMA model selection (p,d,q), forecast values with 95% confidence intervals, model diagnostics, and business interpretation. ### Time Series Decomposition ``` Decompose my quarterly revenue data into trend, seasonal, and irregular components. I want to understand the underlying patterns. Data: quarter,revenue 2020Q1,2400000 2020Q2,1800000 2020Q3,2100000 2020Q4,3200000 2021Q1,2600000 2021Q2,2000000 2021Q3,2300000 2021Q4,3500000 2022Q1,2800000 2022Q2,2200000 2022Q3,2500000 2022Q4,3800000 ``` **Expected Output**: Trend analysis, seasonal patterns, irregular fluctuations, and strategic insights for business planning. ## 📊 Advanced Statistical Testing ### ANOVA with Multiple Comparisons ``` I conducted an experiment testing 4 different marketing strategies on sales performance. Run ANOVA to test for differences and identify which strategies are significantly different. Data: strategy,sales A,1200 A,1180 A,1220 A,1195 A,1210 B,1350 B,1380 B,1340 B,1365 B,1375 C,1180 C,1165 C,1175 C,1170 C,1185 D,1420 D,1445 D,1430 D,1438 D,1452 ``` **Expected Output**: ANOVA F-statistic and p-value, post-hoc pairwise comparisons, effect sizes, and practical recommendations for marketing strategy. ### Chi-Square Test of Independence ``` Test whether customer satisfaction is independent of product category using this survey data. Data: category,satisfaction Electronics,Satisfied Electronics,Very Satisfied Electronics,Neutral Clothing,Satisfied Clothing,Dissatisfied Clothing,Very Satisfied Home,Very Satisfied Home,Satisfied Home,Satisfied Electronics,Very Satisfied Clothing,Neutral Home,Very Satisfied ``` **Expected Output**: Chi-square statistic, p-value, degrees of freedom, Cramér's V effect size, and interpretation of association between variables. ## 🔧 Data Transformation Examples ### Standardization and Outlier Treatment ``` Standardize my financial variables and apply winsorization to handle extreme outliers before modeling. Variables to process: revenue, profit_margin, debt_ratio, roa Data: revenue,profit_margin,debt_ratio,roa 1000000,0.15,0.3,0.08 15000000,0.22,0.4,0.12 500000,0.08,0.8,0.02 8000000,0.18,0.2,0.15 2000000,0.12,0.6,0.05 50000000,0.35,0.1,0.25 800000,0.05,0.9,0.01 ``` **Expected Output**: Standardized variables (mean=0, sd=1), outlier detection report, winsorized values, and recommendations for modeling. ## 🎯 Business Intelligence Scenarios ### Customer Lifetime Value Analysis ``` Calculate CLV using these metrics and segment customers by value tiers. Use clustering to identify distinct customer profiles. Variables: annual_spend, tenure_months, purchase_frequency, avg_order_value, support_tickets Data: customer_id,annual_spend,tenure_months,purchase_frequency,avg_order_value,support_tickets C001,2400,18,24,100,2 C002,6000,36,60,100,1 C003,1200,6,12,100,5 C004,4800,24,24,200,3 C005,9600,48,48,200,0 C006,800,12,8,100,8 C007,7200,30,36,200,2 ``` **Expected Output**: CLV calculations, customer segments (e.g., "High Value", "Growing", "At Risk"), retention strategies, and revenue optimization recommendations. ### Market Basket Analysis ``` Analyze purchase patterns to identify product associations and cross-selling opportunities. Transaction data: transaction_id,products T001,"bread,milk,eggs" T002,"bread,butter,jam" T003,"milk,eggs,cheese" T004,"bread,milk,butter" T005,"eggs,cheese,yogurt" T006,"bread,jam,honey" T007,"milk,cheese,yogurt" ``` **Expected Output**: Association rules (e.g., "If customers buy bread, they're 65% likely to buy milk"), confidence and lift metrics, cross-selling recommendations. ## 💡 Getting Started Tips 1. **Copy-paste any example above** directly into Claude Desktop 2. **Modify the data** to match your real datasets 3. **Ask follow-up questions** like "What if I add more variables?" or "How would this change with more data?" 4. **Request visualizations** by saying "Can you also create plots to show these results?" 5. **Ask for interpretations** with "What do these results mean for my business?" ## 🔍 Advanced Features to Explore - **Model Validation**: "Split this into training/test sets and validate the model" - **Feature Engineering**: "Create interaction terms and polynomial features" - **Ensemble Methods**: "Compare random forest with gradient boosting" - **Bayesian Analysis**: "Run this as a Bayesian regression with priors" - **Survival Analysis**: "Analyze time-to-event data for customer churn" - **Panel Data**: "Control for unobserved heterogeneity with fixed effects" ## 🎯 Pro Tips - Start with simple examples and build complexity - Always ask for business interpretation, not just statistical output - Request confidence intervals and effect sizes for practical significance - Ask follow-up questions to deepen understanding - Combine multiple analyses for comprehensive insights These examples showcase RMCP's comprehensive 429-package ecosystem, demonstrating capabilities far beyond basic statistics through the systematic CRAN task view organization and evidence-based package selection.