Regressor

Regressor is a compact Android app for building, testing, and deploying regression models directly on your device. It combines essential statistical tools—linear, logistic, ridge, and polynomial regression—with interactive charts and CSV import, enabling quick predictive analysis without cloud dependence or a steep learning curve.

Key Features at a Glance

Regressor delivers a focused toolkit for on-device predictive modeling. It supports multiple regression algorithms, parameter tuning, and basic model selection. Users can import datasets in CSV format, configure independent and dependent variables, and obtain coefficient estimates with standard diagnostic metrics such as R-squared, mean squared error, and p-values.

Interface and Workflow

The app is designed for a straightforward workflow: import data, preprocess columns, select a model, and review results. The interface places data preview, variable selection, and algorithm settings within reach of a single screen. Contextual tooltips explain each option, reducing friction for users moving from spreadsheet analysis to statistical modeling.

Model Types and Customization

Regressor includes a practical range of model types suitable for many mobile data tasks. Options include:

• Linear Regression for continuous outcomes with coefficient interpretation and residual plots.
• Logistic Regression for binary classification with probability outputs and confusion matrices.
• Ridge and Lasso regularization to control overfitting on smaller or noisy datasets.
• Polynomial Regression for modeling nonlinear relationships without external preprocessing.

Each model exposes key hyperparameters for manual adjustment and quick comparisons between settings.

Evaluation, Diagnostics, and Validation

Understanding model performance is central to predictive work. Regressor generates diagnostic outputs such as residual distributions, leverage plots, and variance inflation factors where applicable. Built-in cross-validation routines let you split data into training and test sets or run k-fold validation to estimate generalization error. These tools help identify bias, variance, and influential observations before exporting results.

Visualization and Export Options

Visual feedback accelerates insight. The app produces interactive scatter plots, fitted lines, ROC curves for classification, and histograms of residuals. Charts support pinch-to-zoom and value inspection. Results, model parameters, and prediction tables can be exported as CSV files, while graphs are saved as PNG images for reports or presentations.

Performance and Device Compatibility

Designed with mobile constraints in mind, Regressor optimizes memory usage and computation time for typical smartphone datasets. It handles moderate-sized CSV files efficiently and performs numeric operations using native libraries for smoother performance. The app runs on recent Android versions and adapts its layout for phones and tablets, preserving usability across screen sizes.

Security, Privacy, and Offline Use

All processing occurs locally by default, so sensitive datasets remain on the device unless explicitly exported. Regressor requests minimal permissions, primarily file access for importing and exporting data. Offline capability makes it suitable for fieldwork, classroom demonstrations, and situations where connectivity is limited.

Practical Use Cases

• Quick exploratory modeling when away from a laptop or server.
• Teaching regression concepts with hands-on examples and visual diagnostics.
• Field data collection followed by immediate analysis and prediction.
• Preliminary prototyping of models before scaling to larger environments.

Getting Started

Begin by importing a CSV file and reviewing the data preview to identify headers and missing values. Use the variable selector to assign predictors and target variables, then choose a model type. Run a default fit to see core metrics, and iterate by tweaking hyperparameters or trying different preprocessing steps. Export your findings when ready, or save model configurations for later use.

Regressor combines approachable controls with the core statistical outputs practitioners expect, making on-device regression approachable and efficient for analysts, students, and field professionals alike.