Metadata-Version: 2.4
Name: sunlm
Version: 0.1.3.post9
Summary: A social-science-friendly toolkit for OLS regression and SPSS-like ANOVA
Author-email: "Seonwoo (Sun) Kim" <Seonwoo.Kim@nau.edu>
License: MIT
Project-URL: Homepage, https://pypi.org/project/sunlm/
Project-URL: Documentation, https://pypi.org/project/sunlm/
Keywords: statistics,ols,anova,ancova,social science,regression,spss
Requires-Python: >=3.9
Description-Content-Type: text/markdown
Requires-Dist: pandas>=1.3
Requires-Dist: numpy>=1.21
Requires-Dist: statsmodels>=0.13
Requires-Dist: patsy>=0.5

# sunlm

**sunlm** is a lightweight, social-science–friendly Python package for SPSS-like **OLS regression** and **ANOVA / ANCOVA**.  
It is designed for researchers in communication, psychology, marketing, and related social sciences who want **clear, interpretable statistical output** without complex statistical coding.

sunlm emphasizes:
- R-style formula syntax
- SPSS-equivalent ANOVA (Type III SS, sum coding)
- Standardized coefficients and semi-partial R²
- Clean, APA-style tables
- Simple, top-level API (`sunlm.reg`, `sunlm.aov`)

---

## Installation

```bash
pip install sunlm
```

---

## Quick Start: Load Example Data

sunlm includes a built-in example dataset for immediate practice.

```python
import sunlm

df = sunlm.data()
df.head()
```

```text
   personalization  involvement  price_sensitivity  brand_trust        age  purchase_intention
0                1            1          -0.740814     1.408369  51.311519             7.000000
1                0            1           0.072907    -0.608711  41.350777             5.122004
2                0            0           0.403086    -1.320603  27.877310             2.840572
3                1            0           1.471929    -0.669619  46.907773             3.497987
4                1            0           0.307384     1.264625  35.208771             4.472068
```

---

## Tutorial 1️⃣ OLS Regression

```python
import sunlm

model = sunlm.reg(
    "purchase_intention ~ personalization * involvement + age",
    data=df
)

model.summary()
```

```text
## 📊 OLS Regression Summary
                            Unstd. B Std. Err. (unstd) t-value Std. β Semi-partial R²         p-value
Intercept                      2.284             0.473   4.829                         0.000 (< .001)
personalization                0.814             0.319   2.553  0.305           0.035   0.012 (< .05)
involvement                    0.609             0.177   3.443  0.363           0.064  0.001 (< .001)
personalization:involvement    0.374             0.248   1.510  0.204           0.012    0.134 (n.s.)
age                            0.039             0.010   3.713  0.273           0.074  0.000 (< .001)

## 📈 Model Fit Statistics
N            100
R²         0.490
Adj. R²    0.469
```

---

## Tutorial 2️⃣ ANOVA / ANCOVA (SPSS-like)

```python
import sunlm

model = sunlm.aov(
    "purchase_intention ~ C(personalization) * C(involvement) + age",
    data=df
)

model.summary()
```

```text
## Tests of Between-Subjects Effects
Dependent Variable: purchase_intention
                       Source   sum_sq  df mean_sq      F        p-value Partial Eta Squared
              Corrected Model   91.322   6  15.220 16.409 0.000 (< .001)               0.514
                    Intercept   64.597   1  64.597 69.640 0.000 (< .001)               0.428
     personalization (factor)   33.449   1  33.449 36.060 0.000 (< .001)               0.279
         involvement (factor)   40.764   2  20.382 21.973 0.000 (< .001)               0.321
personalization × involvement    2.883   2   1.441  1.554   0.217 (n.s.)               0.032
                          age   11.819   1  11.819 12.742 0.001 (< .001)               0.120
                        Error   86.266  93   0.928                                          
                        Total 2719.617 100                                                  
              Corrected Total  177.588  99                                                  

R Squared = 0.514 (Adjusted R Squared = 0.483)
```

---

## Tutorial 3️⃣ Post Hoc Tests

### Tukey HSD

```python
model.TukeyHSD("involvement")
```

```text
## Tukey HSD for involvement (alpha=0.05)
 group1  group2 meandiff Std. Error        p-value  lower upper
      0       1    1.186      0.235 0.000 (< .001)  0.494 1.877
      0       2    1.494      0.243 0.000 (< .001)  0.779 2.210
      1       2    0.309      0.233   0.534 (n.s.) -0.377 0.995
```

### Bonferroni

```python
model.Bonferroni("involvement")
```

```text
## Bonferroni Pairwise Comparisons for involvement (alpha=0.05)
 group1  group2 meandiff Std. Error     t        p-value  lower upper
      0       1    1.149      0.237 4.846 0.000 (< .001)  0.571 1.727
      0       2    1.576      0.246 6.410 0.000 (< .001)  0.976 2.175
      1       2    0.427      0.234 1.828   0.212 (n.s.) -0.142 0.996
```

---

## Top-level API Summary

```python
sunlm.reg(formula, data)   # OLS regression
sunlm.aov(formula, data)   # ANOVA / ANCOVA
sunlm.data()               # Load example dataset
```

---

## Citation

If you use **sunlm** in academic research, please cite it as:

> Kim, S.-W. (2025). *sunlm* (Version 0.1.3.post1) [Software]. https://pypi.org/project/sunlm/

---

## License

MIT License

---

## Author

**Seonwoo Kim**  
Northern Arizona University  

Designed for researchers who want **SPSS-equivalent statistics with Python clarity**.
