Tasks

• #21
• #22
• #23

Based on concept board

Planning the features and user stories for the evoCampus project.

Possible solutions

• Delete data point
• Replace with aggregate value
  • Mean
  • Median
  • Correlation

(Obj, .vs, ...)

SDS algorithm (uses learning rate, batch size)

Adjust graph limits to the given content

• Requires #57

• 1. Display the input textbox (Ali's dataconverter class)
• 2. Weigths modifier, textbox
• 3. WeightSum (Balazs from Neural class)
• 4. Actvation (Balazs from IActivationFunction)
• 5. Final output.

Create milestones according to planning

Implement error function

• Least Absolute Error
• Least Squared Error

• Extract code duplication

Unit tests for OLS algorithm

• SimpleOrdinaryLeastSquare
• QuadraticOrdinaryLeastSquare

Tasks

• #14
• #15
• #34

Method signature

DataContract(slope, y_intercept)

Add buttons for

• #35
• #36
• #37
• #38

Move #3 to the linear regression application and extend its functionality

Tasks

• #5

Tasks

• #6

After main solutions merge to main branch delete the side branches.

Merge all side branches to main branch

Tasks

DataProvider

• #28
• #29
• #30
• #11

DataModel

• #31
• #32
• #33
• #11

• 1. Neuron class
• 2. Weigthed sum (using Math.Numerics)
• 3. Activation function: tanh, sigmoid, ReLu

Regarding to conceptboard

• 1. Create branch (dataconverter)
• 2. Create WPF Canvas: Frontend
• 3. Converting data: Backend (file 4x4:16)
• 4. Button (calls converting logic)

• Happy path
• All weights zero -> weightSum zero
• Positive, negative weights
• Activation functions (tanh, sigmoid, ReLu)

OLS algorithm (returns m and b)

• - OLS with x^2
• - OLS with mean

Hardcoded m and b

Animate the regression line moving to its trained position using History

• Remove csprojs
• Fix namespaces
• Remove unnecessary checking
• Separate private and public methods
• Simplify Import method

Tasks

• #16
• #17
• #18