PYRSM

Python functions and classes for Business Analytics at the Rady School of Management (RSM), University of California, San Diego (UCSD).

Features

Basics Module - Statistical tests and analyses:

• compare_means - Compare means across groups (t-tests, ANOVA)
• compare_props - Compare proportions between groups
• correlation - Correlation analysis with significance tests
• cross_tabs - Cross-tabulation with chi-square tests
• goodness - Goodness of fit tests
• single_mean - Single sample mean tests
• single_prop - Single sample proportion tests
• prob_calc - Probability calculator for common distributions

Model Module - Regression and machine learning:

• regress - Linear regression with statsmodels
• logistic - Logistic regression with statsmodels
• mlp - Multi-layer perceptron (neural network) with sklearn
• rforest - Random forest with sklearn
• xgboost - XGBoost gradient boosting

EDA Module - Exploratory data analysis:

• explore - Data exploration and summary statistics
• pivot - Pivot tables
• visualize - Data visualization

All modules use Polars DataFrames and plotnine for visualization.

Installation

Requires Python 3.12+ and UV:

mkdir ~/project
cd ~/project
uv init .
uv venv --python 3.13
source .venv/bin/activate
uv add pyrsm

For machine learning models, install with extras:

uv add "pyrsm[ml]"

For all features:

uv add "pyrsm[all]"

If you cloned pyrsm and was to add it in editable format use:

uv add --editable --dev /home/vnijs/gh/pyrsm`

To add to a specific (project) folder use:

uv add --project /opt/base-uv pyrsm`

Quick Start

import polars as pl
import pyrsm as rsmi

# Load data
df = pl.read_parquet("data.parquet")

# Statistical test
cm = rsm.basics.compare_means(df, var="price", byvar="category")
cm.summary()
cm.plot()

# Regression model
reg = rsm.model.regress(df, rvar="price", evar=["size", "age", "type"])
reg.summary()
reg.plot()

R data files to parquet

Convert R data files (.rds, .rda) to parquet files that load into Radiant-AI. Factors become Enums with R's level order, and the markdown description Radiant stores with a dataset is kept inside the parquet file and beside it as <name>_description.md. Requires R (Rscript).

rsm.data.rds_to_parquet("~/data/*.rds")                       # writes ~/data/<name>.parquet
rsm.data.rds_to_parquet("~/data", out_dir="~/parquet")         # every .rds/.rda in a folder
rsm.data.read_description("~/data/diamonds.parquet")

Or from a terminal: rds2parquet ~/data/*.rds (rds2parquet --help for options).

Examples

Extensive example notebooks are available at: https://github.com/radiant-ai-hub/pyrsm/tree/main/examples

License

This project is licensed under the GNU Affero General Public License v3.0.