Class ModelStatsInputs<T, TInput, TOutput>

Namespace

AiDotNet.Models.Inputs

Assembly

AiDotNet.dll

Represents a container for inputs needed to calculate various statistics and metrics for a model.

public class ModelStatsInputs<T, TInput, TOutput>

Type Parameters

T

The numeric type used for calculations, typically float or double.

TInput

The type of input data the model accepts.

TOutput

The type of output data the model produces.

Inheritance

object

ModelStatsInputs<T, TInput, TOutput>

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

This class holds the data and parameters necessary for evaluating model performance and calculating statistics. It includes actual and predicted values, feature information, the model itself, and optional functions for fitting and prediction.

For Beginners: This class collects all the data needed to evaluate how well a model performs.

Think of it like gathering all the ingredients before baking:

• Actual values (what really happened)
• Predicted values (what the model thought would happen)
• Feature data (the information used to make predictions)
• The model itself and its parameters

This organized collection makes it easier to calculate accuracy metrics, perform diagnostic tests, and visualize model performance without having to pass many separate parameters around.

Constructors

ModelStatsInputs()

Initializes a new instance of the ModelStatsInputs<T, TInput, TOutput> class.

public ModelStatsInputs()

Remarks

For Beginners: This creates a new empty container for model statistics inputs.

It initializes all the properties with appropriate empty values based on the types of inputs and outputs your model uses.

Properties

Actual

Gets or sets the actual observed values.

public TOutput Actual { get; set; }

Property Value

TOutput

Remarks

This contains the true values that were observed in reality. These are compared with the model's predictions to evaluate performance and calculate error metrics.

For Beginners: These are the real values that actually occurred.

For example:

• In sales prediction: The actual sales figures that occurred
• In temperature forecasting: The temperatures that were actually recorded
• In classification: The true categories that items belong to

These values serve as the "ground truth" against which we measure how accurate the model's predictions were.

Coefficients

Gets or sets the coefficient values for the model.

public Vector<T> Coefficients { get; set; }

Property Value

Vector<T>

Remarks

This vector contains the coefficients or parameters learned by the model during training. These coefficients determine how the features are used to generate predictions.

For Beginners: These are the weights your model assigned to each feature.

For example, in a house price model:

• A coefficient of 100 for house size means each additional square foot adds $100 to the predicted price
• A coefficient of -5000 for age means each additional year reduces the predicted price by $5000

These values tell you how important each feature is to the model and in what direction (positive or negative) they influence the prediction.

FeatureCount

Gets or sets the number of features or predictor variables in the model.

public int FeatureCount { get; set; }

Property Value

int

Remarks

This property indicates how many features or predictor variables are used in the model. It's important for correctly interpreting the coefficient vector and for certain statistical tests.

For Beginners: This is the number of input variables your model uses to make predictions.

For example:

• A house price model might use 5 features (size, location, age, bedrooms, bathrooms)
• A weather model might use 10 features (temperature, pressure, humidity, etc.)

Knowing the feature count helps interpret model statistics and is needed for certain calculations like degrees of freedom in statistical tests.

FeatureNames

Gets or sets the names of the features used in the model.

public List<string> FeatureNames { get; set; }

Property Value

List<string>

Remarks

This list contains the names of the features or predictor variables used in the model. Having descriptive names makes it easier to interpret coefficients and other statistics.

For Beginners: These are the labels for each input variable your model uses.

For example, in a house price model, feature names might be:

• "SquareFootage"
• "Neighborhood"
• "YearBuilt"
• "Bedrooms"
• "Bathrooms"

Having names instead of just "Feature1", "Feature2", etc. makes it much easier to understand what the model is actually using to make predictions.

FeatureValues

Gets or sets the values for each feature organized by feature name.

public Dictionary<string, TOutput> FeatureValues { get; set; }

Property Value

Dictionary<string, TOutput>

Remarks

This dictionary maps feature names to their values across all observations. The exact structure of the values depends on the model type and data representation. This format can be more convenient than a single input structure for certain operations and visualizations.

For Beginners: This stores all values for each feature separately.

Instead of having one big table, this breaks out each feature into its own collection:

• "SquareFootage": [1500, 2200, 1800, ...]
• "Bedrooms": [3, 4, 2, ...]

This format makes it easier to:

• Calculate statistics for specific features
• Create plots showing individual feature distributions
• Perform feature-specific transformations

FitFunction

Gets or sets a function that fits a model to data.

public Func<TInput, TOutput>? FitFunction { get; set; }

Property Value

Func<TInput, TOutput>

Remarks

This optional function takes input data and target values and returns predicted values. It's used for certain statistical calculations that require refitting the model.

For Beginners: This is a function that can train a model on data.

Think of it as the recipe for how your model learns from data:

• It takes in features (inputs) and actual values
• It figures out the best way to predict the actual values from the features
• It returns predictions based on what it learned

This is used for advanced techniques like cross-validation or bootstrapping where we need to retrain the model multiple times.

Model

Gets or sets the predictive model used to generate the statistics.

public IFullModel<T, TInput, TOutput>? Model { get; set; }

Property Value

IFullModel<T, TInput, TOutput>

Remarks

This property holds a reference to the actual model object that generated the predictions. Having the model available allows for calculating additional statistics or making new predictions.

For Beginners: This is the actual model object that made the predictions.

Having the model itself (not just its predictions) allows us to:

• Make new predictions with different data
• Extract information about how the model works
• Calculate advanced statistics specific to this type of model

It's like having the full recipe book, not just the meal that was cooked.

Predicted

Gets or sets the values predicted by the model.

public TOutput Predicted { get; set; }

Property Value

TOutput

Remarks

This contains the values predicted by the model. These are compared with the actual values to evaluate the model's performance and accuracy.

For Beginners: These are the values that your model predicted would occur.

For example:

• In sales prediction: The sales figures the model predicted
• In temperature forecasting: The temperatures the model forecasted
• In classification: The categories the model assigned to items

By comparing these predictions to what actually happened, we can measure how accurate the model is.

XMatrix

Gets or sets the input data used for predictions.

public TInput XMatrix { get; set; }

Property Value

TInput

Remarks

This contains the feature values used as inputs to the model. The exact structure depends on the model type, but typically each row or element represents an observation with its features.

For Beginners: This is the input data your model used to make predictions.

For example, in a house price prediction model, this might be a table where:

• Each row is a different house
• Each column is a feature (size, location, age, bedrooms, bathrooms)

This data is essential for many advanced statistical calculations, such as confidence intervals and influence measures.