.. _choosing-solution:

====================
Choosing a Solution
====================

NeurEco offers different **solutions** for different types of problems. Each solution contains **templates** optimized for specific use cases. This guide will help you choose the right solution and template for your problem.

Solutions Overview
==================

NeurEco provides three main solutions:

.. list-table::
   :header-rows: 1
   :widths: 20 40 40

   * - Solution
     - Use Case
     - Templates
   * - **Tabular**
     - Static models where inputs and outputs are sets of scalar parameters
     - Regression, Classification, Compression
   * - **Dynamic**
     - Time-series problems where inputs and outputs evolve over time
     - ODE, Convolutional Dynamic
   * - **PFS**
     - Frequency responses of parameterized systems
     - Parametric Frequency Sweep

Decision Flowchart
==================

Use the following flowchart to determine which NeurEco solution and template best fits your problem:

.. figure:: /_images/getting-started/input_data_cnn.png
   :width: 900
   :align: center
   :alt: Decision flowchart for choosing a NeurEco template

   Decision flowchart for choosing the right NeurEco template

Tabular Solution
================

The **Tabular** solution generates static ANN models. Both input and output features are sets of scalar parameters.

**When to use Tabular:**

- Your data is tabular (rows and columns)
- Each sample is independent (no time dependency)
- You want to predict output values from input features

Regression Template
-------------------

Use **Regression** when:

- Your output is a continuous numerical value
- You want to predict quantities like temperature, energy, price, etc.

**Example:** Predicting energy consumption of a house based on temperature, wind speed, humidity, etc.

Classification Template
-----------------------

Use **Classification** when:

- Your output is a category or class label
- You want to classify inputs into discrete groups

**Example:** Classifying cancer types based on RNA gene expression data.

Compression Template
--------------------

Use **Compression** when:

- You want to reduce the dimensionality of your data
- You need to compress and decompress signals
- You're working on IoT applications requiring data reduction

**Example:** Compressing sensor measurements for efficient data transmission.

Dynamic Solution
================

The **Dynamic** solution generates models for time-series data where inputs and outputs evolve over time.

**When to use Dynamic:**

- Your data involves sequences or trajectories
- Time dependencies exist between samples
- You're modeling systems that evolve over time

ODE Template
------------

Use **ODE (Ordinary Differential Equations)** when:

- You're modeling continuous-time dynamical systems
- Time steps are not fixed
- You need to capture differential relationships

**Example:** Modeling physical systems like oscillators, motors, or biological processes.

Convolutional Dynamic Template
------------------------------

Use **Convolutional Dynamic** when:

- You're working with discrete time-series data
- Time steps are fixed
- You need to capture temporal patterns in sequential data

**Example:** Signal processing, sequence prediction, time-series forecasting.

PFS Solution (Parametric Frequency Sweep)
=========================================

The **PFS** solution generates static ANN models for frequency domain problems.

**When to use PFS:**

- Your inputs include scalar parameters plus a frequency
- Your outputs are complex (real and imaginary parts)
- You're modeling frequency responses of systems

**Example:** Modeling the frequency response of electromagnetic systems or mechanical structures.

Quick Reference Table
=====================

.. list-table::
   :header-rows: 1
   :widths: 25 25 50

   * - I need to...
     - Solution/Template
     - Notes
   * - Predict a continuous value
     - Tabular / Regression
     - Most common use case
   * - Classify into categories
     - Tabular / Classification
     - Discrete output labels
   * - Compress/decompress data
     - Tabular / Compression
     - Dimensionality reduction
   * - Model a continuous-time system
     - Dynamic / ODE
     - Variable time steps
   * - Model a discrete-time sequence
     - Dynamic / Convolutional
     - Fixed time steps
   * - Model frequency responses
     - PFS
     - Complex outputs

Still Unsure?
=============

If you're still unsure which solution to use:

1. **Start with Tabular Regression** if your data is tabular and outputs are numerical
2. **Try Dynamic ODE** if your data involves time sequences
3. Contact support@adagos.com for guidance on complex use cases

Next Steps
==========

Once you've identified your solution and template:

1. :ref:`Build your first model <first-model>` using the Energy Consumption example
2. Explore the detailed documentation for your chosen solution in the Solutions Guide