Statsmodels: Statistical Modeling and Econometrics
Overview
Statsmodels is Python's premier library for statistical modeling, providing tools for estimation, inference, and diagnostics across a wide range of statistical methods. Apply this skill for rigorous statistical analysis, from simple linear regression to complex time series models and econometric analyses.
When to Use This Skill
This skill should be used when:
- Fitting regression models (OLS, WLS, GLS, quantile regression)
- Performing generalized linear modeling (logistic, Poisson, Gamma, etc.)
- Analyzing discrete outcomes (binary, multinomial, count, ordinal)
- Conducting time series analysis (ARIMA, SARIMAX, VAR, forecasting)
- Running statistical tests and diagnostics
- Testing model assumptions (heteroskedasticity, autocorrelation, normality)
- Detecting outliers and influential observations
- Comparing models (AIC/BIC, likelihood ratio tests)
- Estimating causal effects
- Producing publication-ready statistical tables and inference
Quick Start Guide
Linear Regression (OLS)
import statsmodels.api as sm
import numpy as np
import pandas as pd
# Prepare data - ALWAYS add constant for intercept
X = sm.add_constant(X_data)
# Fit OLS model
model = sm.OLS(y, X)
results = model.fit()
# View comprehensive results
print(results.summary())
# Key results
print(f"R-squared: {results.rsquared:.4f}")
print(f"Coefficients:\\n{results.params}")
print(f"P-values:\\n{results.pvalues}")
# Predictions with confidence intervals
predictions = results.get_prediction(X_new)
pred_summary = predictions.summary_frame()
print(pred_summary) # includes mean, CI, prediction intervals
# Diagnostics
from statsmodels.stats.diagnostic import het_breuschpagan
bp_test = het_breuschpagan(results.resid, X)
print(f"Breusch-Pagan p-value: {bp_test[1]:.4f}")
# Visualize residuals
import matplotlib.pyplot as plt
plt.scatter(results.fittedvalues, results.resid)
plt.axhline(y=0, color='r', linestyle='--')
plt.xlabel('Fitted values')
plt.ylabel('Residuals')
plt.show()
Logistic Regression (Binary Outcomes)
from statsmodels.discrete.discrete_model import Logit
# Add constant
X = sm.add_constant(X_data)
# Fit logit model
model = Logit(y_binary, X)
results = model.fit()
print(results.summary())
# Odds ratios
odds_ratios = np.exp(results.params)
print("Odds ratios:\\n", odds_ratios)
# Predicted probabilities
probs = results.predict(X)
# Binary predictions (0.5 threshold)
predictions = (probs > 0.5).astype(int)
# Model evaluation
from sklearn.metrics import classification_report, roc_auc_score
print(classification_report(y_binary, predictions))
print(f"AUC: {roc_auc_score(y_binary, probs):.4f}")
# Marginal effects
marginal = results.get_margeff()
print(marginal.summary())
Time Series (ARIMA)
from statsmodels.tsa.arima.model import ARIMA
from statsmodels.graphics.tsaplots import plot_acf, plot_pacf
# Check stationarity
from statsmodels.tsa.stattools import adfuller
adf_result = adfuller(y_series)
print(f"ADF p-value: {adf_result[1]:.4f}")
if adf_result[1] > 0.05:
# Series is non-stationary, difference it
y_diff = y_series.diff().dropna()
# Plot ACF/PACF to identify p, q
fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(12, 8))
plot_acf(y_diff, lags=40, ax=ax1)
plot_pacf(y_diff, lags=40, ax=ax2)
plt.show()
# Fit ARIMA(p,d,q)
model = ARIMA(y_series, order=(1, 1, 1))
results = model.fit()
print(results.summary())
# Forecast
forecast = results.forecast(steps=10)
forecast_obj = results.get_forecast(steps=10)
forecast_df = forecast_obj.summary_frame()
print(forecast_df) # includes mean and confidence intervals
# Residual diagnostics
results.plot_diagnostics(figsize=(12, 8))
plt.show()
Generalized Linear Models (GLM)
import statsmodels.api as sm
# Poisson regression for count data
X = sm.add_constant(X_data)
model = sm.GLM(y_counts, X, family=sm.families.Poisson())
results = model.fit()
print(results.summary())
# Rate ratios (for Poisson with log link)
rate_ratios = np.exp(results.params)
print("Rate ratios:\\n", rate_ratios)
# Check overdispersion
overdispersion = results.pearson_chi2 / results.df_resid
print(f"Overdispersion: {overdispersion:.2f}")
if overdispersion > 1.5:
# Use Negative Binomial instead
from statsmodels.discrete.count_model import NegativeBinomial
nb_model = NegativeBinomial(y_counts, X)
nb_results = nb_model.fit()
print(nb_results.summary())
Core Statistical Modeling Capabilities
1. Linear Regression Models
Comprehensive suite of linear models for continuous outcomes with various error structures.
Available models:
- OLS: Standard linear regression with i.i.d. errors
- WLS: Weighted least squares for heteroskedastic errors
- GLS: Generalized least squares for arbitrary covariance structure
- GLSAR: GLS with autoregressive errors for time series
- Quantile Regression: Conditional quantiles (robust to outliers)
- Mixed Effects: Hierarchical/multilevel models with random effects
- Recursive/Rolling: Time-varying parameter estimation
Key features:
- Comprehensive diagnostic tests
- Robust standard errors (HC, HAC, cluster-robust)
- Influence statistics (Cook's distance, leverage, DFFITS)
- Hypothesis testing (F-tests, Wald tests)
- Model comparison (AIC, BIC, likelihood ratio tests)
- Prediction with confidence and prediction intervals
When to use: Continuous outcome variable, want inference on coefficients, need diagnostics
Reference: See references/linear_models.md for detailed guidance on model selection, diagnostics, and best practices.
2. Generalized Linear Models (GLM)
Flexible framework extending linear models to non-normal distributions.
Distribution families:
- Binomial: Binary outcomes or proportions (logistic regression)
- Poisson: Count data
- Negative Binomial: Overdispersed counts
- Gamma: Positive continuous, right-skewed data
- Inverse Gaussian: Positive continuous with specific variance structure
- Gaussian: Equivalent to OLS
- Tweedie: Flexible family for semi-continuous data
Link functions:
- Logit, Probit, Log, Identity, Inverse, Sqrt, CLogLog, Power
- Choose based on interpretation needs and model fit
Key features:
- Maximum likelihood estimation via IRLS
- Deviance and Pearson residuals
- Goodness-of-fit statistics
- Pseudo R-squared measures
- Robust standard errors
When to use: Non-normal outcomes, need flexible variance and link specifications
Reference: See references/glm.md for family selection, link functions, interpretation, and diagnostics.
3. Discrete Choice Models
Models for categorical and count outcomes.
Binary models:
- Logit: Logistic regression (odds ratios)
- Probit: Probit regression (normal distribution)
Multinomial models:
- MNLogit: Unordered categories (3+ levels)
- Conditional Logit: Choice models with alternative-specific variables
- Ordered Model: Ordinal outcomes (ordered categories)
Count models:
- Poisson: Standard count model
- Negative Binomial: Overdispersed counts
- Zero-Inflated: Excess zeros (ZIP, ZINB)
- Hurdle Models: Two-stage models for zero-heavy data
Key features:
- Maximum likelihood estimation
- Marginal effects at means or average marginal effects
- Model comparison via AIC/BIC
- Predicted probabilities and classification
- Goodness-of-fit tests
When to use: Binary, categorical, or count outcomes
Reference: See references/discrete_choice.md for model selection, interpretation, and evaluation.
4. Time Series Analysis
Comprehensive time series modeling and forecasting capabilities.
Univariate models:
- AutoReg (AR): Autoregressive models
- ARIMA: Autoregressive integrated moving average
- SARIMAX: Seasonal ARIMA with exogenous variables
- Exponential Smoothing: Simple, Holt, Holt-Winters
- ETS: Innovations state space models
Multivariate models:
- VAR: Vector autoregression
- VARMAX: VAR with MA and exogenous variables
- Dynamic Factor Models: Extract common factors
- VECM: Vector error correction models (cointegration)
Advanced models:
- State Space: Kalman filtering, custom specifications
- Regime Switching: Markov switching models
- ARDL: Autoregressive distributed lag
Key features:
- ACF/PACF analysis for model identification
- Stationarity tests (ADF, KPSS)
- Forecasting with prediction intervals
- Residual diagnostics (Ljung-Box, heteroskedasticity)
- Granger causality testing
- Impulse response functions (IRF)
- Forecast error variance decomposition (FEVD)
When to use: Time-ordered data, forecasting, understanding temporal dynamics
Reference: See references/time_series.md for model selection, diagnostics, and forecasting methods.
5. Statistical Tests and Diagnostics
Extensive testing and diagnostic capabilities for model validation.
Residual diagnostics:
- Autocorrelation tests (Ljung-Box, Durbin-Watson, Breusch-Godfrey)
- Heteroskedasticity tests (Breusch-Pagan, White, ARCH)
- Normality tests (Jarque-Bera, Omnibus, Anderson-Darling, Lilliefors)
- Specification tests (RESET, Harvey-Collier)
Influence and outliers:
- Leverage (hat values)
- Cook's distance
- DFFITS and DFBETAs
- Studentized residuals
- Influence plots
Hypothesis testing:
- t-tests (one-sample, two-sample, paired)
- Proportion tests
- Chi-square tests
- Non-parametric tests (Mann-Whitney, Wilcoxon, Kruskal-Wallis)
- ANOVA (one-way, two-way, repeated measures)
Multiple comparisons:
- Tukey's HSD
- Bonferroni correction
- False Discovery Rate (FDR)
Effect sizes and power:
- Cohen's d, eta-squared
- Power analysis for t-tests, proportions
- Sample size calculations
Robust inference:
- Heteroskedasticity-consistent SEs (HC0-HC3)
- HAC standard errors (Newey-West)
- Cluster-robust standard errors
When to use: Validating assumptions, detecting problems, ensuring robust inference
Reference: See references/stats_diagnostics.md for comprehensive testing and diagnostic procedures.
Formula API (R-style)
Statsmodels supports R-style formulas for intuitive model specification:
import statsmodels.formula.api as smf
# OLS with formula
results = smf.ols('y ~ x1 + x2 + x1:x2', data=df).fit()
# Categorical variables (automatic dummy coding)
results = smf.ols('y ~ x1 + C(category)', data=df).fit()
# Interactions
results = smf.ols('y ~ x1 * x2', data=df).fit() # x1 + x2 + x1:x2
# Polynomial terms
results = smf.ols('y ~ x + I(x**2)', data=df).fit()
# Logit
results = smf.logit('y ~ x1 + x2 + C(group)', data=df).fit()
# Poisson
results = smf.poisson('count ~ x1 + x2', data=df).fit()
# ARIMA (not available via formula, use regular API)
Model Selection and Comparison
Information Criteria
# Compare models using AIC/BIC
models = {
'Model 1': model1_results,
'Model 2': model2_results,
'Model 3': model3_results
}
comparison = pd.DataFrame({
'AIC': {name: res.aic for name, res in models.items()},
'BIC': {name: res.bic for name, res in models.items()},
'Log-Likelihood': {name: res.llf for name, res in models.items()}
})
print(comparison.sort_values('AIC'))
# Lower AIC/BIC indicates better model
Likelihood Ratio Test (Nested Models)
# For nested models (one is subset of the other)
from scipy import stats
lr_stat = 2 * (full_model.llf - reduced_model.llf)
df = full_model.df_model - reduced_model.df_model
p_value = 1 - stats.chi2.cdf(lr_stat, df)
print(f"LR statistic: {lr_stat:.4f}")
print(f"p-value: {p_value:.4f}")
if p_value < 0.05:
print("Full model significantly better")
else:
print("Reduced model preferred (parsimony)")
Cross-Validation
from sklearn.model_selection import KFold
from sklearn.metrics import mean_squared_error
kf = KFold(n_splits=5, shuffle=True, random_state=42)
cv_scores = []
for train_idx, val_idx in kf.split(X):
X_train, X_val = X.iloc[train_idx], X.iloc[val_idx]
y_train, y_val = y.iloc[train_idx], y.iloc[val_idx]
# Fit model
model = sm.OLS(y_train, X_train).fit()
# Predict
y_pred = model.predict(X_val)
# Score
rmse = np.sqrt(mean_squared_error(y_val, y_pred))
cv_scores.append(rmse)
print(f"CV RMSE: {np.mean(cv_scores):.4f} ± {np.std(cv_scores):.4f}")
Best Practices
Data Preparation
- Always add constant: Use
sm.add_constant() unless excluding intercept
- Check for missing values: Handle or impute before fitting
- Scale if needed: Improves convergence, interpretation (but not required for tree models)
- Encode categoricals: Use formula API or manual dummy coding
Model Building
- Start simple: Begin with basic model, add complexity as needed
- Check assumptions: Test residuals, heteroskedasticity, autocorrelation
- Use appropriate model: Match model to outcome type (binary→Logit, count→Poisson)
- Consider alternatives: If assumptions violated, use robust methods or different model
Inference
- Report effect sizes: Not just p-values
- Use robust SEs: When heteroskedasticity or clustering present
- Multiple comparisons: Correct when testing many hypotheses
- Confidence intervals: Always report alongside point estimates
Model Evaluation
- Check residuals: Plot residuals vs fitted, Q-Q plot
- Influence diagnostics: Identify and investigate influential observations
- Out-of-sample validation: Test on holdout set or cross-validate
- Compare models: Use AIC/BIC for non-nested, LR test for nested
Reporting
- Comprehensive summary: Use
.summary() for detailed output
- Document decisions: Note transformations, excluded observations
- Interpret carefully: Account for link functions (e.g., exp(β) for log link)
- Visualize: Plot predictions, confidence intervals, diagnostics
Common Workflows
Workflow 1: Linear Regression Analysis
- Explore data (plots, descriptives)
- Fit initial OLS model
- Check residual diagnostics
- Test for heteroskedasticity, autocorrelation
- Check for multicollinearity (VIF)
- Identify influential observations
- Refit with robust SEs if needed
- Interpret coefficients and inference
- Validate on holdout or via CV
Workflow 2: Binary Classification
- Fit logistic regression (Logit)
- Check for convergence issues
- Interpret odds ratios
- Calculate marginal effects
- Evaluate classification performance (AUC, confusion matrix)
- Check for influential observations
- Compare with alternative models (Probit)
- Validate predictions on test set
Workflow 3: Count Data Analysis
- Fit Poisson regression
- Check for overdispersion
- If overdispersed, fit Negative Binomial
- Check for excess zeros (consider ZIP/ZINB)
- Interpret rate ratios
- Assess goodness of fit
- Compare models via AIC
- Validate predictions
Workflow 4: Time Series Forecasting
- Plot series, check for trend/seasonality
- Test for stationarity (ADF, KPSS)
- Difference if non-stationary
- Identify p, q from ACF/PACF
- Fit ARIMA or SARIMAX
- Check residual diagnostics (Ljung-Box)
- Generate forecasts with confidence intervals
- Evaluate forecast accuracy on test set
Reference Documentation
This skill includes comprehensive reference files for detailed guidance:
references/linear_models.md
Detailed coverage of linear regression models including:
- OLS, WLS, GLS, GLSAR, Quantile Regression
- Mixed effects models
- Recursive and rolling regression
- Comprehensive diagnostics (heteroskedasticity, autocorrelation, multicollinearity)
- Influence statistics and outlier detection
- Robust standard errors (HC, HAC, cluster)
- Hypothesis testing and model comparison
references/glm.md
Complete guide to generalized linear models:
- All distribution families (Binomial, Poisson, Gamma, etc.)
- Link functions and when to use each
- Model fitting and interpretation
- Pseudo R-squared and goodness of fit
- Diagnostics and residual analysis
- Applications (logistic, Poisson, Gamma regression)
references/discrete_choice.md
Comprehensive guide to discrete outcome models:
- Binary models (Logit, Probit)
- Multinomial models (MNLogit, Conditional Logit)
- Count models (Poisson, Negative Binomial, Zero-Inflated, Hurdle)
- Ordinal models
- Marginal effects and interpretation
- Model diagnostics and comparison
references/time_series.md
In-depth time series analysis guidance:
- Univariate models (AR, ARIMA, SARIMAX, Exponential Smoothing)
- Multivariate models (VAR, VARMAX, Dynamic Factor)
- State space models
- Stationarity testing and diagnostics
- Forecasting methods and evaluation
- Granger causality, IRF, FEVD
references/stats_diagnostics.md
Comprehensive statistical testing and diagnostics:
- Residual diagnostics (autocorrelation, heteroskedasticity, normality)
- Influence and outlier detection
- Hypothesis tests (parametric and non-parametric)
- ANOVA and post-hoc tests
- Multiple comparisons correction
- Robust covariance matrices
- Power analysis and effect sizes
When to reference:
- Need detailed parameter explanations
- Choosing between similar models
- Troubleshooting convergence or diagnostic issues
- Understanding specific test statistics
- Looking for code examples for advanced features
Search patterns:
# Find information about specific models
grep -r "Quantile Regression" references/
# Find diagnostic tests
grep -r "Breusch-Pagan" references/stats_diagnostics.md
# Find time series guidance
grep -r "SARIMAX" references/time_series.md
Common Pitfalls to Avoid
- Forgetting constant term: Always use
sm.add_constant() unless no intercept desired
- Ignoring assumptions: Check residuals, heteroskedasticity, autocorrelation
- Wrong model for outcome type: Binary→Logit/Probit, Count→Poisson/NB, not OLS
- Not checking convergence: Look for optimization warnings
- Misinterpreting coefficients: Remember link functions (log, logit, etc.)
- Using Poisson with overdispersion: Check dispersion, use Negative Binomial if needed
- Not using robust SEs: When heteroskedasticity or clustering present
- Overfitting: Too many parameters relative to sample size
- Data leakage: Fitting on test data or using future information
- Not validating predictions: Always check out-of-sample performance
- Comparing non-nested models: Use AIC/BIC, not LR test
- Ignoring influential observations: Check Cook's distance and leverage
- Multiple testing: Correct p-values when testing many hypotheses
- Not differencing time series: Fit ARIMA on non-stationary data
- Confusing prediction vs confidence intervals: Prediction intervals are wider
Getting Help
For detailed documentation and examples:
1---2name: statsmodels3description: Statistical models library for Python. Use when you need specific model classes (OLS, GLM, mixed models, ARIMA) with detailed diagnostics, residuals, and inference. Best for econometrics, time series, rigorous inference with coefficient tables. For guided statistical test selection with APA reporting use statistical-analysis.4category: coding5license: BSD-3-Clause license6metadata:7 source: https://skillmd.com/skills/synthetic-sciences/statsmodels8 upstream: K-Dense-AI/scientific-agent-skills9 upstream-url: https://github.com/K-Dense-AI/scientific-agent-skills10 upstream-path: skills/statsmodels11 upstream-license: MIT12 upstream-relationship: derived13 adapted-by: Synthetic Sciences14 skill-author: K-Dense Inc.15version: 1.0.016author: K-Dense Inc.17tags: [Statistics, Econometrics, Time Series, Regression]18dependencies: ["statsmodels>=0.14.0", "numpy>=1.25.0", "scipy>=1.12.0"]19---2021# Statsmodels: Statistical Modeling and Econometrics2223## Overview2425Statsmodels is Python's premier library for statistical modeling, providing tools for estimation, inference, and diagnostics across a wide range of statistical methods. Apply this skill for rigorous statistical analysis, from simple linear regression to complex time series models and econometric analyses.2627## When to Use This Skill2829This skill should be used when:30- Fitting regression models (OLS, WLS, GLS, quantile regression)31- Performing generalized linear modeling (logistic, Poisson, Gamma, etc.)32- Analyzing discrete outcomes (binary, multinomial, count, ordinal)33- Conducting time series analysis (ARIMA, SARIMAX, VAR, forecasting)34- Running statistical tests and diagnostics35- Testing model assumptions (heteroskedasticity, autocorrelation, normality)36- Detecting outliers and influential observations37- Comparing models (AIC/BIC, likelihood ratio tests)38- Estimating causal effects39- Producing publication-ready statistical tables and inference4041## Quick Start Guide4243### Linear Regression (OLS)4445```python46import statsmodels.api as sm47import numpy as np48import pandas as pd4950# Prepare data - ALWAYS add constant for intercept51X = sm.add_constant(X_data)5253# Fit OLS model54model = sm.OLS(y, X)55results = model.fit()5657# View comprehensive results58print(results.summary())5960# Key results61print(f"R-squared: {results.rsquared:.4f}")62print(f"Coefficients:\\n{results.params}")63print(f"P-values:\\n{results.pvalues}")6465# Predictions with confidence intervals66predictions = results.get_prediction(X_new)67pred_summary = predictions.summary_frame()68print(pred_summary) # includes mean, CI, prediction intervals6970# Diagnostics71from statsmodels.stats.diagnostic import het_breuschpagan72bp_test = het_breuschpagan(results.resid, X)73print(f"Breusch-Pagan p-value: {bp_test[1]:.4f}")7475# Visualize residuals76import matplotlib.pyplot as plt77plt.scatter(results.fittedvalues, results.resid)78plt.axhline(y=0, color='r', linestyle='--')79plt.xlabel('Fitted values')80plt.ylabel('Residuals')81plt.show()82```8384### Logistic Regression (Binary Outcomes)8586```python87from statsmodels.discrete.discrete_model import Logit8889# Add constant90X = sm.add_constant(X_data)9192# Fit logit model93model = Logit(y_binary, X)94results = model.fit()9596print(results.summary())9798# Odds ratios99odds_ratios = np.exp(results.params)100print("Odds ratios:\\n", odds_ratios)101102# Predicted probabilities103probs = results.predict(X)104105# Binary predictions (0.5 threshold)106predictions = (probs > 0.5).astype(int)107108# Model evaluation109from sklearn.metrics import classification_report, roc_auc_score110111print(classification_report(y_binary, predictions))112print(f"AUC: {roc_auc_score(y_binary, probs):.4f}")113114# Marginal effects115marginal = results.get_margeff()116print(marginal.summary())117```118119### Time Series (ARIMA)120121```python122from statsmodels.tsa.arima.model import ARIMA123from statsmodels.graphics.tsaplots import plot_acf, plot_pacf124125# Check stationarity126from statsmodels.tsa.stattools import adfuller127128adf_result = adfuller(y_series)129print(f"ADF p-value: {adf_result[1]:.4f}")130131if adf_result[1] > 0.05:132 # Series is non-stationary, difference it133 y_diff = y_series.diff().dropna()134135# Plot ACF/PACF to identify p, q136fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(12, 8))137plot_acf(y_diff, lags=40, ax=ax1)138plot_pacf(y_diff, lags=40, ax=ax2)139plt.show()140141# Fit ARIMA(p,d,q)142model = ARIMA(y_series, order=(1, 1, 1))143results = model.fit()144145print(results.summary())146147# Forecast148forecast = results.forecast(steps=10)149forecast_obj = results.get_forecast(steps=10)150forecast_df = forecast_obj.summary_frame()151152print(forecast_df) # includes mean and confidence intervals153154# Residual diagnostics155results.plot_diagnostics(figsize=(12, 8))156plt.show()157```158159### Generalized Linear Models (GLM)160161```python162import statsmodels.api as sm163164# Poisson regression for count data165X = sm.add_constant(X_data)166model = sm.GLM(y_counts, X, family=sm.families.Poisson())167results = model.fit()168169print(results.summary())170171# Rate ratios (for Poisson with log link)172rate_ratios = np.exp(results.params)173print("Rate ratios:\\n", rate_ratios)174175# Check overdispersion176overdispersion = results.pearson_chi2 / results.df_resid177print(f"Overdispersion: {overdispersion:.2f}")178179if overdispersion > 1.5:180 # Use Negative Binomial instead181 from statsmodels.discrete.count_model import NegativeBinomial182 nb_model = NegativeBinomial(y_counts, X)183 nb_results = nb_model.fit()184 print(nb_results.summary())185```186187## Core Statistical Modeling Capabilities188189### 1. Linear Regression Models190191Comprehensive suite of linear models for continuous outcomes with various error structures.192193**Available models:**194- **OLS**: Standard linear regression with i.i.d. errors195- **WLS**: Weighted least squares for heteroskedastic errors196- **GLS**: Generalized least squares for arbitrary covariance structure197- **GLSAR**: GLS with autoregressive errors for time series198- **Quantile Regression**: Conditional quantiles (robust to outliers)199- **Mixed Effects**: Hierarchical/multilevel models with random effects200- **Recursive/Rolling**: Time-varying parameter estimation201202**Key features:**203- Comprehensive diagnostic tests204- Robust standard errors (HC, HAC, cluster-robust)205- Influence statistics (Cook's distance, leverage, DFFITS)206- Hypothesis testing (F-tests, Wald tests)207- Model comparison (AIC, BIC, likelihood ratio tests)208- Prediction with confidence and prediction intervals209210**When to use:** Continuous outcome variable, want inference on coefficients, need diagnostics211212**Reference:** See `references/linear_models.md` for detailed guidance on model selection, diagnostics, and best practices.213214### 2. Generalized Linear Models (GLM)215216Flexible framework extending linear models to non-normal distributions.217218**Distribution families:**219- **Binomial**: Binary outcomes or proportions (logistic regression)220- **Poisson**: Count data221- **Negative Binomial**: Overdispersed counts222- **Gamma**: Positive continuous, right-skewed data223- **Inverse Gaussian**: Positive continuous with specific variance structure224- **Gaussian**: Equivalent to OLS225- **Tweedie**: Flexible family for semi-continuous data226227**Link functions:**228- Logit, Probit, Log, Identity, Inverse, Sqrt, CLogLog, Power229- Choose based on interpretation needs and model fit230231**Key features:**232- Maximum likelihood estimation via IRLS233- Deviance and Pearson residuals234- Goodness-of-fit statistics235- Pseudo R-squared measures236- Robust standard errors237238**When to use:** Non-normal outcomes, need flexible variance and link specifications239240**Reference:** See `references/glm.md` for family selection, link functions, interpretation, and diagnostics.241242### 3. Discrete Choice Models243244Models for categorical and count outcomes.245246**Binary models:**247- **Logit**: Logistic regression (odds ratios)248- **Probit**: Probit regression (normal distribution)249250**Multinomial models:**251- **MNLogit**: Unordered categories (3+ levels)252- **Conditional Logit**: Choice models with alternative-specific variables253- **Ordered Model**: Ordinal outcomes (ordered categories)254255**Count models:**256- **Poisson**: Standard count model257- **Negative Binomial**: Overdispersed counts258- **Zero-Inflated**: Excess zeros (ZIP, ZINB)259- **Hurdle Models**: Two-stage models for zero-heavy data260261**Key features:**262- Maximum likelihood estimation263- Marginal effects at means or average marginal effects264- Model comparison via AIC/BIC265- Predicted probabilities and classification266- Goodness-of-fit tests267268**When to use:** Binary, categorical, or count outcomes269270**Reference:** See `references/discrete_choice.md` for model selection, interpretation, and evaluation.271272### 4. Time Series Analysis273274Comprehensive time series modeling and forecasting capabilities.275276**Univariate models:**277- **AutoReg (AR)**: Autoregressive models278- **ARIMA**: Autoregressive integrated moving average279- **SARIMAX**: Seasonal ARIMA with exogenous variables280- **Exponential Smoothing**: Simple, Holt, Holt-Winters281- **ETS**: Innovations state space models282283**Multivariate models:**284- **VAR**: Vector autoregression285- **VARMAX**: VAR with MA and exogenous variables286- **Dynamic Factor Models**: Extract common factors287- **VECM**: Vector error correction models (cointegration)288289**Advanced models:**290- **State Space**: Kalman filtering, custom specifications291- **Regime Switching**: Markov switching models292- **ARDL**: Autoregressive distributed lag293294**Key features:**295- ACF/PACF analysis for model identification296- Stationarity tests (ADF, KPSS)297- Forecasting with prediction intervals298- Residual diagnostics (Ljung-Box, heteroskedasticity)299- Granger causality testing300- Impulse response functions (IRF)301- Forecast error variance decomposition (FEVD)302303**When to use:** Time-ordered data, forecasting, understanding temporal dynamics304305**Reference:** See `references/time_series.md` for model selection, diagnostics, and forecasting methods.306307### 5. Statistical Tests and Diagnostics308309Extensive testing and diagnostic capabilities for model validation.310311**Residual diagnostics:**312- Autocorrelation tests (Ljung-Box, Durbin-Watson, Breusch-Godfrey)313- Heteroskedasticity tests (Breusch-Pagan, White, ARCH)314- Normality tests (Jarque-Bera, Omnibus, Anderson-Darling, Lilliefors)315- Specification tests (RESET, Harvey-Collier)316317**Influence and outliers:**318- Leverage (hat values)319- Cook's distance320- DFFITS and DFBETAs321- Studentized residuals322- Influence plots323324**Hypothesis testing:**325- t-tests (one-sample, two-sample, paired)326- Proportion tests327- Chi-square tests328- Non-parametric tests (Mann-Whitney, Wilcoxon, Kruskal-Wallis)329- ANOVA (one-way, two-way, repeated measures)330331**Multiple comparisons:**332- Tukey's HSD333- Bonferroni correction334- False Discovery Rate (FDR)335336**Effect sizes and power:**337- Cohen's d, eta-squared338- Power analysis for t-tests, proportions339- Sample size calculations340341**Robust inference:**342- Heteroskedasticity-consistent SEs (HC0-HC3)343- HAC standard errors (Newey-West)344- Cluster-robust standard errors345346**When to use:** Validating assumptions, detecting problems, ensuring robust inference347348**Reference:** See `references/stats_diagnostics.md` for comprehensive testing and diagnostic procedures.349350## Formula API (R-style)351352Statsmodels supports R-style formulas for intuitive model specification:353354```python355import statsmodels.formula.api as smf356357# OLS with formula358results = smf.ols('y ~ x1 + x2 + x1:x2', data=df).fit()359360# Categorical variables (automatic dummy coding)361results = smf.ols('y ~ x1 + C(category)', data=df).fit()362363# Interactions364results = smf.ols('y ~ x1 * x2', data=df).fit() # x1 + x2 + x1:x2365366# Polynomial terms367results = smf.ols('y ~ x + I(x**2)', data=df).fit()368369# Logit370results = smf.logit('y ~ x1 + x2 + C(group)', data=df).fit()371372# Poisson373results = smf.poisson('count ~ x1 + x2', data=df).fit()374375# ARIMA (not available via formula, use regular API)376```377378## Model Selection and Comparison379380### Information Criteria381382```python383# Compare models using AIC/BIC384models = {385 'Model 1': model1_results,386 'Model 2': model2_results,387 'Model 3': model3_results388}389390comparison = pd.DataFrame({391 'AIC': {name: res.aic for name, res in models.items()},392 'BIC': {name: res.bic for name, res in models.items()},393 'Log-Likelihood': {name: res.llf for name, res in models.items()}394})395396print(comparison.sort_values('AIC'))397# Lower AIC/BIC indicates better model398```399400### Likelihood Ratio Test (Nested Models)401402```python403# For nested models (one is subset of the other)404from scipy import stats405406lr_stat = 2 * (full_model.llf - reduced_model.llf)407df = full_model.df_model - reduced_model.df_model408p_value = 1 - stats.chi2.cdf(lr_stat, df)409410print(f"LR statistic: {lr_stat:.4f}")411print(f"p-value: {p_value:.4f}")412413if p_value < 0.05:414 print("Full model significantly better")415else:416 print("Reduced model preferred (parsimony)")417```418419### Cross-Validation420421```python422from sklearn.model_selection import KFold423from sklearn.metrics import mean_squared_error424425kf = KFold(n_splits=5, shuffle=True, random_state=42)426cv_scores = []427428for train_idx, val_idx in kf.split(X):429 X_train, X_val = X.iloc[train_idx], X.iloc[val_idx]430 y_train, y_val = y.iloc[train_idx], y.iloc[val_idx]431432 # Fit model433 model = sm.OLS(y_train, X_train).fit()434435 # Predict436 y_pred = model.predict(X_val)437438 # Score439 rmse = np.sqrt(mean_squared_error(y_val, y_pred))440 cv_scores.append(rmse)441442print(f"CV RMSE: {np.mean(cv_scores):.4f} ± {np.std(cv_scores):.4f}")443```444445## Best Practices446447### Data Preparation4484491. **Always add constant**: Use `sm.add_constant()` unless excluding intercept4502. **Check for missing values**: Handle or impute before fitting4513. **Scale if needed**: Improves convergence, interpretation (but not required for tree models)4524. **Encode categoricals**: Use formula API or manual dummy coding453454### Model Building4554561. **Start simple**: Begin with basic model, add complexity as needed4572. **Check assumptions**: Test residuals, heteroskedasticity, autocorrelation4583. **Use appropriate model**: Match model to outcome type (binary→Logit, count→Poisson)4594. **Consider alternatives**: If assumptions violated, use robust methods or different model460461### Inference4624631. **Report effect sizes**: Not just p-values4642. **Use robust SEs**: When heteroskedasticity or clustering present4653. **Multiple comparisons**: Correct when testing many hypotheses4664. **Confidence intervals**: Always report alongside point estimates467468### Model Evaluation4694701. **Check residuals**: Plot residuals vs fitted, Q-Q plot4712. **Influence diagnostics**: Identify and investigate influential observations4723. **Out-of-sample validation**: Test on holdout set or cross-validate4734. **Compare models**: Use AIC/BIC for non-nested, LR test for nested474475### Reporting4764771. **Comprehensive summary**: Use `.summary()` for detailed output4782. **Document decisions**: Note transformations, excluded observations4793. **Interpret carefully**: Account for link functions (e.g., exp(β) for log link)4804. **Visualize**: Plot predictions, confidence intervals, diagnostics481482## Common Workflows483484### Workflow 1: Linear Regression Analysis4854861. Explore data (plots, descriptives)4872. Fit initial OLS model4883. Check residual diagnostics4894. Test for heteroskedasticity, autocorrelation4905. Check for multicollinearity (VIF)4916. Identify influential observations4927. Refit with robust SEs if needed4938. Interpret coefficients and inference4949. Validate on holdout or via CV495496### Workflow 2: Binary Classification4974981. Fit logistic regression (Logit)4992. Check for convergence issues5003. Interpret odds ratios5014. Calculate marginal effects5025. Evaluate classification performance (AUC, confusion matrix)5036. Check for influential observations5047. Compare with alternative models (Probit)5058. Validate predictions on test set506507### Workflow 3: Count Data Analysis5085091. Fit Poisson regression5102. Check for overdispersion5113. If overdispersed, fit Negative Binomial5124. Check for excess zeros (consider ZIP/ZINB)5135. Interpret rate ratios5146. Assess goodness of fit5157. Compare models via AIC5168. Validate predictions517518### Workflow 4: Time Series Forecasting5195201. Plot series, check for trend/seasonality5212. Test for stationarity (ADF, KPSS)5223. Difference if non-stationary5234. Identify p, q from ACF/PACF5245. Fit ARIMA or SARIMAX5256. Check residual diagnostics (Ljung-Box)5267. Generate forecasts with confidence intervals5278. Evaluate forecast accuracy on test set528529## Reference Documentation530531This skill includes comprehensive reference files for detailed guidance:532533### references/linear_models.md534Detailed coverage of linear regression models including:535- OLS, WLS, GLS, GLSAR, Quantile Regression536- Mixed effects models537- Recursive and rolling regression538- Comprehensive diagnostics (heteroskedasticity, autocorrelation, multicollinearity)539- Influence statistics and outlier detection540- Robust standard errors (HC, HAC, cluster)541- Hypothesis testing and model comparison542543### references/glm.md544Complete guide to generalized linear models:545- All distribution families (Binomial, Poisson, Gamma, etc.)546- Link functions and when to use each547- Model fitting and interpretation548- Pseudo R-squared and goodness of fit549- Diagnostics and residual analysis550- Applications (logistic, Poisson, Gamma regression)551552### references/discrete_choice.md553Comprehensive guide to discrete outcome models:554- Binary models (Logit, Probit)555- Multinomial models (MNLogit, Conditional Logit)556- Count models (Poisson, Negative Binomial, Zero-Inflated, Hurdle)557- Ordinal models558- Marginal effects and interpretation559- Model diagnostics and comparison560561### references/time_series.md562In-depth time series analysis guidance:563- Univariate models (AR, ARIMA, SARIMAX, Exponential Smoothing)564- Multivariate models (VAR, VARMAX, Dynamic Factor)565- State space models566- Stationarity testing and diagnostics567- Forecasting methods and evaluation568- Granger causality, IRF, FEVD569570### references/stats_diagnostics.md571Comprehensive statistical testing and diagnostics:572- Residual diagnostics (autocorrelation, heteroskedasticity, normality)573- Influence and outlier detection574- Hypothesis tests (parametric and non-parametric)575- ANOVA and post-hoc tests576- Multiple comparisons correction577- Robust covariance matrices578- Power analysis and effect sizes579580**When to reference:**581- Need detailed parameter explanations582- Choosing between similar models583- Troubleshooting convergence or diagnostic issues584- Understanding specific test statistics585- Looking for code examples for advanced features586587**Search patterns:**588```bash589# Find information about specific models590grep -r "Quantile Regression" references/591592# Find diagnostic tests593grep -r "Breusch-Pagan" references/stats_diagnostics.md594595# Find time series guidance596grep -r "SARIMAX" references/time_series.md597```598599## Common Pitfalls to Avoid6006011. **Forgetting constant term**: Always use `sm.add_constant()` unless no intercept desired6022. **Ignoring assumptions**: Check residuals, heteroskedasticity, autocorrelation6033. **Wrong model for outcome type**: Binary→Logit/Probit, Count→Poisson/NB, not OLS6044. **Not checking convergence**: Look for optimization warnings6055. **Misinterpreting coefficients**: Remember link functions (log, logit, etc.)6066. **Using Poisson with overdispersion**: Check dispersion, use Negative Binomial if needed6077. **Not using robust SEs**: When heteroskedasticity or clustering present6088. **Overfitting**: Too many parameters relative to sample size6099. **Data leakage**: Fitting on test data or using future information61010. **Not validating predictions**: Always check out-of-sample performance61111. **Comparing non-nested models**: Use AIC/BIC, not LR test61212. **Ignoring influential observations**: Check Cook's distance and leverage61313. **Multiple testing**: Correct p-values when testing many hypotheses61414. **Not differencing time series**: Fit ARIMA on non-stationary data61515. **Confusing prediction vs confidence intervals**: Prediction intervals are wider616617## Getting Help618619For detailed documentation and examples:620- Official docs: https://www.statsmodels.org/stable/621- User guide: https://www.statsmodels.org/stable/user-guide.html622- Examples: https://www.statsmodels.org/stable/examples/index.html623- API reference: https://www.statsmodels.org/stable/api.html624