packages & analytics

Descriptive

uses current and historical data to describe trends and relationships.

• communications change over time effectively

EX] “what happened?”

descriptive

the objective is to get a fundamental understanding of your data

Diagnostic

determines the root cause of trends and correlation between variables

• crucial for understanding the factors contributing to a given outcome

EX] “why did this happen?”

diagnostic

the objective is to dig deeper into the data to determine why something happened.

Predictive

uses historical data to forecast future scenarios, trends, an events to inform business decisions.

EX] “what might happen in the future?”

Prescriptive

forecast future outcomes, PLUS recommend actions to benefit from predictions.

EX] “What should we do next?”

prescriptive

the objective is to consider various possible decisions and identify the best course of action.

recipe

data preprocessing

parsnip

model specification

Workflows

streamlining model fitting

tune

hyperparameter optimization

yardstick

model evaluation

broom

tidying model outputs

tidymodels

a collection of R packages that provides a comprehensive framework, designed to work seamlessly within the tidyverse ecosystem.

• recipes

• tune

• parsnip

• yardstick

• rsample

rsample

provides infrastructure for efficient data splitting and resampling.

Yardstick

measures the effectiveness of models using performance metrics.

Broom

converts the information in common statistical R objects into user-friendly, predictable formats

dials

creates and manages tuning parameters and parameter grids.

workflows

provides a cohesive framework that binds that binds together and preprocessing steps (recipes) and model specifications (parsnip) into a single, unified object.

key advantages of using workflows

• unified process

• reproducibility

• flexibility and efficiency

key applications in data modeling

1. Forecasting future trends

2. Optimizing business strategies

3. Enhancing decision-making

4. Driving innovation and discovery

5. Understanding complex patterns

6. Strategic edge

key aspects in predictive modeling

1. Feature engineering

2. Model interpretability

3. Transparency and trust

4. Handling imbalanced data

5. Ethical considerations and fairness

6. Model deployment and monitoring

7. Collaboration and communication

Recipes

allows to define the model formula and specify preprocessing steps to the original dataset