plottah 1.3.2

Plottah: Univariate Analysis

Copyright © 2024 by Boston Consulting Group. All rights reserved

SmartBanking Plotting tool

A Python package for generating standardized univariate analysis plots and visualizations for SmartBanking analysis. The package automatically generates:

• ROC curves
• Distribution plots
• Bin event rate plots

Installation

There are two ways to use this package:

1. Install from PyPI (Recommended)

If you just want to use the package as a library:

pip install plottah

2. Local Development with Config (For Batch Processing)

If you want to process multiple features using a configuration file:

1. Clone the repository:

git clone [e-mail]:nielsota/plottah.git
cd plottah

2. Install Poetry (if you haven't already):

curl -sSL https://install.python-poetry.org | python3 -

3. Install dependencies and set up the development environment:

poetry install

Usage

As a Python Package

from plottah import build_univariate_plot
import pandas as pd

# Create your dataframe
df = pd.DataFrame(...)

# Generate a single plot
plot = build_univariate_plot(
    df=df,
    feature_col="your_feature",
    target="target_column",
    feature_type="numerical",  # or "categorical"
    n_bins=10,  # optional
    distplot_q_min=0.01,  # optional
    distplot_q_max=0.99   # optional
)

Using Configuration File

1. Create a config.yaml file with your settings:

file_path: ./data/your_data.csv
images_output_path: ./data/images
powerpoint_output_path: ./data/powerpoints/output.pptx

features:
  - name: feature_1
    n_bins: 10
  - name: feature_2
    bins: [0, 1, 5, 10, 100]
  - name: feature_3
    type: categorical

target: target_column

# Optional: Custom colors
primary_color: 231, 30, 87
secondary_color: 153, 204, 235
tertiary_color: 254, 189, 64
grey_tint_color: 110, 111, 115

2. Run the plotting tool:

poetry run python -m plottah

Configuration Options

For each feature in the config file, you can specify:

• name: Feature column name (required)
• type: "numerical" or "categorical" (default: "numerical")
• n_bins: Number of bins for numerical features
• bins: Custom bin edges for numerical features
• distplot_q_min: Lower quantile for distribution plot trimming
• distplot_q_max: Upper quantile for distribution plot trimming

Examples

See the notebooks/ directory for detailed examples of:

• Basic univariate analysis
• Custom binning strategies
• Distribution plot customization
• ROC curve analysis

Requirements

• Python >=3.8
• Poetry for development setup