You are on page 1of 8

Mauro Sebastián Innocente

PhD, MSc, MEng, FHEA

Autonomous Vehicles &


Artificial Intelligence Laboratory (AVAILab)
Mauro.S.Innocente@gmail.com
https://msinnocente.com/
https://availab.org/

AVAILab
Autonomous Vehicles &
Artificial Intelligence
Laboratory
• Nonlinear Regression
• Nonlinear Regression using Simple Linear Regression
• Polynomial Regression using Multiple Linear Regression

Nonlinear Regression Mauro S. Innocente


2|8
• Nonlinear Regression
• Nonlinear Regression using Simple Linear Regression
• Polynomial Regression using Multiple Linear Regression

Nonlinear Regression Mauro S. Innocente


3|8
• Observed data modelled by a nonlinear function of the model parameters.
• Nonlinearity arises from the fact that the approximation function (hypothesis)
cannot be expressed as a linear combination of the parameters of the model.
• Examples: exponential, logarithmic, trigonometric, and power functions.
• The optimisation problem becomes more difficult. Training Set

• Example: Consider the following data. 0


x
0
y

• By observing a plot of the raw data, it is 10 77


estimated that the approximation function 20 118
30 132
may be an exponential one. 40 145
• Hence a Two-Term Exponential Model 50 149
60 148
is fitted (4 unknown coefficients).
70 153
𝑔(𝑥) = 𝑎 ⋅ 𝑒 𝑏⋅𝑥 + 𝑐 ⋅ 𝑒 𝑑⋅𝑥 80 150
90 154
• Solving the optimisation problem: 100 155
𝑔(𝑥) = 148.7587 ⋅ 𝑒 0.0004⋅𝑥 − 149.0275 ⋅ 𝑒 −0.0741⋅𝑥
Nonlinear Regression Mauro S. Innocente
4|8
• Nonlinear Regression
• Nonlinear Regression using Simple Linear Regression
• Polynomial Regression using Multiple Linear Regression

Nonlinear Regression Mauro S. Innocente


5|8
𝑔(𝑥) = 𝑎 ⋅ 𝑒 𝑏⋅𝑥

ln 𝑔(𝑥) = ln 𝑎 + 𝑏 ⋅ 𝑥

𝑔′(𝑥) = 𝑎0 + 𝑎1 ⋅ 𝑥

𝑔(𝑥) = 𝑎 ⋅ 𝑥 𝑏

ln 𝑔(𝑥) = ln 𝑎 + 𝑏 ⋅ ln 𝑥

𝑔′(𝑥) = 𝑎0 + 𝑎1 ⋅ 𝑥′

𝑎 = 𝑒 𝑎0

𝑏 = 𝑎1

Nonlinear Regression Mauro S. Innocente


6|8
• Nonlinear Regression
• Nonlinear Regression using Simple Linear Regression
• Polynomial Regression using Multiple Linear Regression

Nonlinear Regression Mauro S. Innocente


7|8
𝑔(𝑥) = 𝑎0 + 𝑎1 ⋅ 𝑥 + 𝑎2 ⋅ 𝑥 2

𝑔(𝑥) = 𝑎0 + 𝑎1 ⋅ 𝑥1 + 𝑎2 ⋅ 𝑥2

Nonlinear Regression Mauro S. Innocente


8|8

You might also like