Formulas in T

Formulas provide a declarative way to specify statistical models, inspired by R.

Syntax

response ~ predictor

The ~ operator creates a Formula object that can be passed to modeling functions.

Examples

-- Simple linear regression
model = lm(data = df, formula = y ~ x)

-- Multiple regression
model = lm(data = df, formula = y ~ x1 + x2 + x3)

-- Interaction terms
model = lm(data = df, formula = y ~ x1 * x2)

Supported Functions

• lm() - Linear regression

lm() - Linear Regression

Fit a linear model using least squares.

Signature

lm(data: DataFrame, formula: Formula, ...) -> Dict

Arguments

• data: DataFrame containing the variables
• formula: Formula specifying the model (e.g., y ~ x)

Returns

Dictionary containing:

• formula: The model formula
• coefficients: Dictionary of term estimates
• std_errors: Dictionary of standard errors
• r_squared: R² statistic
• adj_r_squared: Adjusted R² statistic
• sigma: Residual standard error
• nobs: Number of observations
• _tidy_df: Tidy DataFrame of results

Examples

data = read_csv("mtcars.csv")
model = lm(data = data, formula = mpg ~ hp)
print(model.r_squared)

Current Limitations

• Intercept control: y ~ x + 1 vs y ~ x - 1
• Transformations: y ~ log(x)

---

Next Steps

Now that you’ve explored formulas, learn about statistical modeling and pipelines in T:

1. Models and Prediction — Comprehensive guide to statistical models.
2. Pipeline Tutorial — Build reproducible data pipelines.
3. API Reference — Complete function reference by package.