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: "statsmodels"3description: "Statistical modeling toolkit. OLS, GLM, logistic, ARIMA, time series, hypothesis tests, diagnostics, AIC/BIC, for rigorous statistical inference and econometric analysis."4metadata:5 source: https://skillmd.com/skills/viggyv/statsmodels6---78# Statsmodels: Statistical Modeling and Econometrics910## Overview1112Statsmodels 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.1314## When to Use This Skill1516This skill should be used when:17- Fitting regression models (OLS, WLS, GLS, quantile regression)18- Performing generalized linear modeling (logistic, Poisson, Gamma, etc.)19- Analyzing discrete outcomes (binary, multinomial, count, ordinal)20- Conducting time series analysis (ARIMA, SARIMAX, VAR, forecasting)21- Running statistical tests and diagnostics22- Testing model assumptions (heteroskedasticity, autocorrelation, normality)23- Detecting outliers and influential observations24- Comparing models (AIC/BIC, likelihood ratio tests)25- Estimating causal effects26- Producing publication-ready statistical tables and inference2728## Quick Start Guide2930### Linear Regression (OLS)3132```python33import statsmodels.api as sm34import numpy as np35import pandas as pd3637# Prepare data - ALWAYS add constant for intercept38X = sm.add_constant(X_data)3940# Fit OLS model41model = sm.OLS(y, X)42results = model.fit()4344# View comprehensive results45print(results.summary())4647# Key results48print(f"R-squared: {results.rsquared:.4f}")49print(f"Coefficients:\\n{results.params}")50print(f"P-values:\\n{results.pvalues}")5152# Predictions with confidence intervals53predictions = results.get_prediction(X_new)54pred_summary = predictions.summary_frame()55print(pred_summary) # includes mean, CI, prediction intervals5657# Diagnostics58from statsmodels.stats.diagnostic import het_breuschpagan59bp_test = het_breuschpagan(results.resid, X)60print(f"Breusch-Pagan p-value: {bp_test[1]:.4f}")6162# Visualize residuals63import matplotlib.pyplot as plt64plt.scatter(results.fittedvalues, results.resid)65plt.axhline(y=0, color='r', linestyle='--')66plt.xlabel('Fitted values')67plt.ylabel('Residuals')68plt.show()69```7071### Logistic Regression (Binary Outcomes)7273```python74from statsmodels.discrete.discrete_model import Logit7576# Add constant77X = sm.add_constant(X_data)7879# Fit logit model80model = Logit(y_binary, X)81results = model.fit()8283print(results.summary())8485# Odds ratios86odds_ratios = np.exp(results.params)87print("Odds ratios:\\n", odds_ratios)8889# Predicted probabilities90probs = results.predict(X)9192# Binary predictions (0.5 threshold)93predictions = (probs > 0.5).astype(int)9495# Model evaluation96from sklearn.metrics import classification_report, roc_auc_score9798print(classification_report(y_binary, predictions))99print(f"AUC: {roc_auc_score(y_binary, probs):.4f}")100101# Marginal effects102marginal = results.get_margeff()103print(marginal.summary())104```105106### Time Series (ARIMA)107108```python109from statsmodels.tsa.arima.model import ARIMA110from statsmodels.graphics.tsaplots import plot_acf, plot_pacf111112# Check stationarity113from statsmodels.tsa.stattools import adfuller114115adf_result = adfuller(y_series)116print(f"ADF p-value: {adf_result[1]:.4f}")117118if adf_result[1] > 0.05:119 # Series is non-stationary, difference it120 y_diff = y_series.diff().dropna()121122# Plot ACF/PACF to identify p, q123fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(12, 8))124plot_acf(y_diff, lags=40, ax=ax1)125plot_pacf(y_diff, lags=40, ax=ax2)126plt.show()127128# Fit ARIMA(p,d,q)129model = ARIMA(y_series, order=(1, 1, 1))130results = model.fit()131132print(results.summary())133134# Forecast135forecast = results.forecast(steps=10)136forecast_obj = results.get_forecast(steps=10)137forecast_df = forecast_obj.summary_frame()138139print(forecast_df) # includes mean and confidence intervals140141# Residual diagnostics142results.plot_diagnostics(figsize=(12, 8))143plt.show()144```145146### Generalized Linear Models (GLM)147148```python149import statsmodels.api as sm150151# Poisson regression for count data152X = sm.add_constant(X_data)153model = sm.GLM(y_counts, X, family=sm.families.Poisson())154results = model.fit()155156print(results.summary())157158# Rate ratios (for Poisson with log link)159rate_ratios = np.exp(results.params)160print("Rate ratios:\\n", rate_ratios)161162# Check overdispersion163overdispersion = results.pearson_chi2 / results.df_resid164print(f"Overdispersion: {overdispersion:.2f}")165166if overdispersion > 1.5:167 # Use Negative Binomial instead168 from statsmodels.discrete.count_model import NegativeBinomial169 nb_model = NegativeBinomial(y_counts, X)170 nb_results = nb_model.fit()171 print(nb_results.summary())172```173174## Core Statistical Modeling Capabilities175176### 1. Linear Regression Models177178Comprehensive suite of linear models for continuous outcomes with various error structures.179180**Available models:**181- **OLS**: Standard linear regression with i.i.d. errors182- **WLS**: Weighted least squares for heteroskedastic errors183- **GLS**: Generalized least squares for arbitrary covariance structure184- **GLSAR**: GLS with autoregressive errors for time series185- **Quantile Regression**: Conditional quantiles (robust to outliers)186- **Mixed Effects**: Hierarchical/multilevel models with random effects187- **Recursive/Rolling**: Time-varying parameter estimation188189**Key features:**190- Comprehensive diagnostic tests191- Robust standard errors (HC, HAC, cluster-robust)192- Influence statistics (Cook's distance, leverage, DFFITS)193- Hypothesis testing (F-tests, Wald tests)194- Model comparison (AIC, BIC, likelihood ratio tests)195- Prediction with confidence and prediction intervals196197**When to use:** Continuous outcome variable, want inference on coefficients, need diagnostics198199**Reference:** See `references/linear_models.md` for detailed guidance on model selection, diagnostics, and best practices.200201### 2. Generalized Linear Models (GLM)202203Flexible framework extending linear models to non-normal distributions.204205**Distribution families:**206- **Binomial**: Binary outcomes or proportions (logistic regression)207- **Poisson**: Count data208- **Negative Binomial**: Overdispersed counts209- **Gamma**: Positive continuous, right-skewed data210- **Inverse Gaussian**: Positive continuous with specific variance structure211- **Gaussian**: Equivalent to OLS212- **Tweedie**: Flexible family for semi-continuous data213214**Link functions:**215- Logit, Probit, Log, Identity, Inverse, Sqrt, CLogLog, Power216- Choose based on interpretation needs and model fit217218**Key features:**219- Maximum likelihood estimation via IRLS220- Deviance and Pearson residuals221- Goodness-of-fit statistics222- Pseudo R-squared measures223- Robust standard errors224225**When to use:** Non-normal outcomes, need flexible variance and link specifications226227**Reference:** See `references/glm.md` for family selection, link functions, interpretation, and diagnostics.228229### 3. Discrete Choice Models230231Models for categorical and count outcomes.232233**Binary models:**234- **Logit**: Logistic regression (odds ratios)235- **Probit**: Probit regression (normal distribution)236237**Multinomial models:**238- **MNLogit**: Unordered categories (3+ levels)239- **Conditional Logit**: Choice models with alternative-specific variables240- **Ordered Model**: Ordinal outcomes (ordered categories)241242**Count models:**243- **Poisson**: Standard count model244- **Negative Binomial**: Overdispersed counts245- **Zero-Inflated**: Excess zeros (ZIP, ZINB)246- **Hurdle Models**: Two-stage models for zero-heavy data247248**Key features:**249- Maximum likelihood estimation250- Marginal effects at means or average marginal effects251- Model comparison via AIC/BIC252- Predicted probabilities and classification253- Goodness-of-fit tests254255**When to use:** Binary, categorical, or count outcomes256257**Reference:** See `references/discrete_choice.md` for model selection, interpretation, and evaluation.258259### 4. Time Series Analysis260261Comprehensive time series modeling and forecasting capabilities.262263**Univariate models:**264- **AutoReg (AR)**: Autoregressive models265- **ARIMA**: Autoregressive integrated moving average266- **SARIMAX**: Seasonal ARIMA with exogenous variables267- **Exponential Smoothing**: Simple, Holt, Holt-Winters268- **ETS**: Innovations state space models269270**Multivariate models:**271- **VAR**: Vector autoregression272- **VARMAX**: VAR with MA and exogenous variables273- **Dynamic Factor Models**: Extract common factors274- **VECM**: Vector error correction models (cointegration)275276**Advanced models:**277- **State Space**: Kalman filtering, custom specifications278- **Regime Switching**: Markov switching models279- **ARDL**: Autoregressive distributed lag280281**Key features:**282- ACF/PACF analysis for model identification283- Stationarity tests (ADF, KPSS)284- Forecasting with prediction intervals285- Residual diagnostics (Ljung-Box, heteroskedasticity)286- Granger causality testing287- Impulse response functions (IRF)288- Forecast error variance decomposition (FEVD)289290**When to use:** Time-ordered data, forecasting, understanding temporal dynamics291292**Reference:** See `references/time_series.md` for model selection, diagnostics, and forecasting methods.293294### 5. Statistical Tests and Diagnostics295296Extensive testing and diagnostic capabilities for model validation.297298**Residual diagnostics:**299- Autocorrelation tests (Ljung-Box, Durbin-Watson, Breusch-Godfrey)300- Heteroskedasticity tests (Breusch-Pagan, White, ARCH)301- Normality tests (Jarque-Bera, Omnibus, Anderson-Darling, Lilliefors)302- Specification tests (RESET, Harvey-Collier)303304**Influence and outliers:**305- Leverage (hat values)306- Cook's distance307- DFFITS and DFBETAs308- Studentized residuals309- Influence plots310311**Hypothesis testing:**312- t-tests (one-sample, two-sample, paired)313- Proportion tests314- Chi-square tests315- Non-parametric tests (Mann-Whitney, Wilcoxon, Kruskal-Wallis)316- ANOVA (one-way, two-way, repeated measures)317318**Multiple comparisons:**319- Tukey's HSD320- Bonferroni correction321- False Discovery Rate (FDR)322323**Effect sizes and power:**324- Cohen's d, eta-squared325- Power analysis for t-tests, proportions326- Sample size calculations327328**Robust inference:**329- Heteroskedasticity-consistent SEs (HC0-HC3)330- HAC standard errors (Newey-West)331- Cluster-robust standard errors332333**When to use:** Validating assumptions, detecting problems, ensuring robust inference334335**Reference:** See `references/stats_diagnostics.md` for comprehensive testing and diagnostic procedures.336337## Formula API (R-style)338339Statsmodels supports R-style formulas for intuitive model specification:340341```python342import statsmodels.formula.api as smf343344# OLS with formula345results = smf.ols('y ~ x1 + x2 + x1:x2', data=df).fit()346347# Categorical variables (automatic dummy coding)348results = smf.ols('y ~ x1 + C(category)', data=df).fit()349350# Interactions351results = smf.ols('y ~ x1 * x2', data=df).fit() # x1 + x2 + x1:x2352353# Polynomial terms354results = smf.ols('y ~ x + I(x**2)', data=df).fit()355356# Logit357results = smf.logit('y ~ x1 + x2 + C(group)', data=df).fit()358359# Poisson360results = smf.poisson('count ~ x1 + x2', data=df).fit()361362# ARIMA (not available via formula, use regular API)363```364365## Model Selection and Comparison366367### Information Criteria368369```python370# Compare models using AIC/BIC371models = {372 'Model 1': model1_results,373 'Model 2': model2_results,374 'Model 3': model3_results375}376377comparison = pd.DataFrame({378 'AIC': {name: res.aic for name, res in models.items()},379 'BIC': {name: res.bic for name, res in models.items()},380 'Log-Likelihood': {name: res.llf for name, res in models.items()}381})382383print(comparison.sort_values('AIC'))384# Lower AIC/BIC indicates better model385```386387### Likelihood Ratio Test (Nested Models)388389```python390# For nested models (one is subset of the other)391from scipy import stats392393lr_stat = 2 * (full_model.llf - reduced_model.llf)394df = full_model.df_model - reduced_model.df_model395p_value = 1 - stats.chi2.cdf(lr_stat, df)396397print(f"LR statistic: {lr_stat:.4f}")398print(f"p-value: {p_value:.4f}")399400if p_value < 0.05:401 print("Full model significantly better")402else:403 print("Reduced model preferred (parsimony)")404```405406### Cross-Validation407408```python409from sklearn.model_selection import KFold410from sklearn.metrics import mean_squared_error411412kf = KFold(n_splits=5, shuffle=True, random_state=42)413cv_scores = []414415for train_idx, val_idx in kf.split(X):416 X_train, X_val = X.iloc[train_idx], X.iloc[val_idx]417 y_train, y_val = y.iloc[train_idx], y.iloc[val_idx]418419 # Fit model420 model = sm.OLS(y_train, X_train).fit()421422 # Predict423 y_pred = model.predict(X_val)424425 # Score426 rmse = np.sqrt(mean_squared_error(y_val, y_pred))427 cv_scores.append(rmse)428429print(f"CV RMSE: {np.mean(cv_scores):.4f} ± {np.std(cv_scores):.4f}")430```431432## Best Practices433434### Data Preparation4354361. **Always add constant**: Use `sm.add_constant()` unless excluding intercept4372. **Check for missing values**: Handle or impute before fitting4383. **Scale if needed**: Improves convergence, interpretation (but not required for tree models)4394. **Encode categoricals**: Use formula API or manual dummy coding440441### Model Building4424431. **Start simple**: Begin with basic model, add complexity as needed4442. **Check assumptions**: Test residuals, heteroskedasticity, autocorrelation4453. **Use appropriate model**: Match model to outcome type (binary→Logit, count→Poisson)4464. **Consider alternatives**: If assumptions violated, use robust methods or different model447448### Inference4494501. **Report effect sizes**: Not just p-values4512. **Use robust SEs**: When heteroskedasticity or clustering present4523. **Multiple comparisons**: Correct when testing many hypotheses4534. **Confidence intervals**: Always report alongside point estimates454455### Model Evaluation4564571. **Check residuals**: Plot residuals vs fitted, Q-Q plot4582. **Influence diagnostics**: Identify and investigate influential observations4593. **Out-of-sample validation**: Test on holdout set or cross-validate4604. **Compare models**: Use AIC/BIC for non-nested, LR test for nested461462### Reporting4634641. **Comprehensive summary**: Use `.summary()` for detailed output4652. **Document decisions**: Note transformations, excluded observations4663. **Interpret carefully**: Account for link functions (e.g., exp(β) for log link)4674. **Visualize**: Plot predictions, confidence intervals, diagnostics468469## Common Workflows470471### Workflow 1: Linear Regression Analysis4724731. Explore data (plots, descriptives)4742. Fit initial OLS model4753. Check residual diagnostics4764. Test for heteroskedasticity, autocorrelation4775. Check for multicollinearity (VIF)4786. Identify influential observations4797. Refit with robust SEs if needed4808. Interpret coefficients and inference4819. Validate on holdout or via CV482483### Workflow 2: Binary Classification4844851. Fit logistic regression (Logit)4862. Check for convergence issues4873. Interpret odds ratios4884. Calculate marginal effects4895. Evaluate classification performance (AUC, confusion matrix)4906. Check for influential observations4917. Compare with alternative models (Probit)4928. Validate predictions on test set493494### Workflow 3: Count Data Analysis4954961. Fit Poisson regression4972. Check for overdispersion4983. If overdispersed, fit Negative Binomial4994. Check for excess zeros (consider ZIP/ZINB)5005. Interpret rate ratios5016. Assess goodness of fit5027. Compare models via AIC5038. Validate predictions504505### Workflow 4: Time Series Forecasting5065071. Plot series, check for trend/seasonality5082. Test for stationarity (ADF, KPSS)5093. Difference if non-stationary5104. Identify p, q from ACF/PACF5115. Fit ARIMA or SARIMAX5126. Check residual diagnostics (Ljung-Box)5137. Generate forecasts with confidence intervals5148. Evaluate forecast accuracy on test set515516## Reference Documentation517518This skill includes comprehensive reference files for detailed guidance:519520### references/linear_models.md521Detailed coverage of linear regression models including:522- OLS, WLS, GLS, GLSAR, Quantile Regression523- Mixed effects models524- Recursive and rolling regression525- Comprehensive diagnostics (heteroskedasticity, autocorrelation, multicollinearity)526- Influence statistics and outlier detection527- Robust standard errors (HC, HAC, cluster)528- Hypothesis testing and model comparison529530### references/glm.md531Complete guide to generalized linear models:532- All distribution families (Binomial, Poisson, Gamma, etc.)533- Link functions and when to use each534- Model fitting and interpretation535- Pseudo R-squared and goodness of fit536- Diagnostics and residual analysis537- Applications (logistic, Poisson, Gamma regression)538539### references/discrete_choice.md540Comprehensive guide to discrete outcome models:541- Binary models (Logit, Probit)542- Multinomial models (MNLogit, Conditional Logit)543- Count models (Poisson, Negative Binomial, Zero-Inflated, Hurdle)544- Ordinal models545- Marginal effects and interpretation546- Model diagnostics and comparison547548### references/time_series.md549In-depth time series analysis guidance:550- Univariate models (AR, ARIMA, SARIMAX, Exponential Smoothing)551- Multivariate models (VAR, VARMAX, Dynamic Factor)552- State space models553- Stationarity testing and diagnostics554- Forecasting methods and evaluation555- Granger causality, IRF, FEVD556557### references/stats_diagnostics.md558Comprehensive statistical testing and diagnostics:559- Residual diagnostics (autocorrelation, heteroskedasticity, normality)560- Influence and outlier detection561- Hypothesis tests (parametric and non-parametric)562- ANOVA and post-hoc tests563- Multiple comparisons correction564- Robust covariance matrices565- Power analysis and effect sizes566567**When to reference:**568- Need detailed parameter explanations569- Choosing between similar models570- Troubleshooting convergence or diagnostic issues571- Understanding specific test statistics572- Looking for code examples for advanced features573574**Search patterns:**575```bash576# Find information about specific models577grep -r "Quantile Regression" references/578579# Find diagnostic tests580grep -r "Breusch-Pagan" references/stats_diagnostics.md581582# Find time series guidance583grep -r "SARIMAX" references/time_series.md584```585586## Common Pitfalls to Avoid5875881. **Forgetting constant term**: Always use `sm.add_constant()` unless no intercept desired5892. **Ignoring assumptions**: Check residuals, heteroskedasticity, autocorrelation5903. **Wrong model for outcome type**: Binary→Logit/Probit, Count→Poisson/NB, not OLS5914. **Not checking convergence**: Look for optimization warnings5925. **Misinterpreting coefficients**: Remember link functions (log, logit, etc.)5936. **Using Poisson with overdispersion**: Check dispersion, use Negative Binomial if needed5947. **Not using robust SEs**: When heteroskedasticity or clustering present5958. **Overfitting**: Too many parameters relative to sample size5969. **Data leakage**: Fitting on test data or using future information59710. **Not validating predictions**: Always check out-of-sample performance59811. **Comparing non-nested models**: Use AIC/BIC, not LR test59912. **Ignoring influential observations**: Check Cook's distance and leverage60013. **Multiple testing**: Correct p-values when testing many hypotheses60114. **Not differencing time series**: Fit ARIMA on non-stationary data60215. **Confusing prediction vs confidence intervals**: Prediction intervals are wider603604## Getting Help605606For detailed documentation and examples:607- Official docs: https://www.statsmodels.org/stable/608- User guide: https://www.statsmodels.org/stable/user-guide.html609- Examples: https://www.statsmodels.org/stable/examples/index.html610- API reference: https://www.statsmodels.org/stable/api.html