Constructor for the RidgeThreadedReg Object which uses unweighted ridge regression to correlate click and mouse movement to eye patch features
This class has the same functionality as RidgeReg with two large differences:
1. training examples are batched for retraining the model instead of retraining on each new example
2. the training runs on a separate thread which should enable better running time
- Source:
Methods
addData(eyes, screenPos, type)
adds data to the regression model
Parameters:
| Name | Type | Description |
|---|---|---|
eyes |
object | util.eyes Object containing left and right data |
screenPos |
array | the screen [x,y] position when a training event happens |
type |
string | the type of event |
- Source:
getData() → {Array.<Object>}
gets the training data stored in this regression model, *this is not the model itself, but merely its training data*
- Source:
Returns:
the set of training data stored in this regression class
- Type
- Array.<Object>
predict(eyesObj) → {Object|integer|integer}
gets a prediction based on the current set of training data
Parameters:
| Name | Type | Description |
|---|---|---|
eyesObj |
object | util.eyes Object |
- Source:
Returns:
-
prediction - Object containing the prediction data
- Type
- Object
-
prediction.x - the x screen coordinate predicted
- Type
- integer
-
prediction.y - the y screen coordinate predicted
- Type
- integer
setData()
seeds the model with initial training data in case data is stored in a separate location
- Source: