srishti013 / time_series_classification

Time Series Classification

• In this project I have developed a Time Series classification model based on Fourier Transform. I have used three classifiers Random Forest, Logistic Regression and Nearest Neighbour. After that I have compared their performances using scatter plots.

• The datasets used to make the models are present here

Results

• Results

• These results show the error values each model has with the corresponding dataset.

• The three classifiers used are:

• Nearest Neighbour(ed)
• Random Forest(rf)
• Logistic Regression(lr)

• All three classifier are used with the following three values derived from fourier transform of orignal dataset:

• Phase and Amplitude(phase_amp)
• Real and Imaginary values(real_imag)
• Phase, Amplitude, Real and Imaginary values(real_imag_phase_amp)

Wins,Losses And Ties (With Nearest Neighbour(DTW))

• These models are used on fourier transform of the given dataset.

1) Nearest neighbour ( Euclidean Distance - Real, Imaginary, Phase and Amplitude)

• Wins=61
  Tie=2
  Loss=50
  

2) Nearest neighbour ( Euclidean Distance - Phase and Amplitude)

• Wins=55
  Tie=1
  Loss=57
  

3) Nearest neighbour ( Euclidean Distance - Real and Imaginary)

• Wins=62
  Tie=1
  Loss=50
  

4) Logistic Regression ( Real and Imaginary)

• Wins=37
  Tie=4
  Loss=72
  

5) Logistic Regression ( Real, Imaginary, Phase and Amplitude)

• Wins=48
  Tie=4
  Loss=61
  

6) Logistic Regression ( Phase and Amplitude)

• Wins=36
  Tie=2
  Loss=75
  

7) Random Forest ( Phase and Amplitude)

• Wins=54
  Tie=3
  Loss=56
  

8) Random Forest (Real and Imaginary)

• Wins=40
  Tie=3
  Loss=70
  

9) Random Forest (Real, Imaginary, Phase and Amplitude)

• Wins=53
  Tie=2
  Loss=58
  

Scatter plots

1) Random Forest(1)

2) Random Forest(2)

3) Random Forest(3)

4) Logistic Regression(1)

5) Logistic Regression(2)

6) Logistic Regression(3)

7) Nearest Neighbour(1)

8) Nearest Neighbour(2)

9) Nearest Neighbour(3)

10) Nearest Neighbour(DTW) VS Nearest Neighbour(ED)(1)

11) Nearest Neighbour(DTW) VS Nearest Neighbour(ED)(2)

12) Nearest Neighbour(DTW) VS Nearest Neighbour(ED)(3)

13) Logistic Regression Vs Neighourest Neighbour(DTW)(1)

14) Logistic Regression Vs Neighourest Neighbour(DTW)(2)

15) Logistic Regression Vs Neighourest Neighbour(DTW)(3)

16) Random Forest Vs Neighourest Neighbour(DTW)(1)

17) Random Forest Vs Neighourest Neighbour(DTW)(2)

18) Random Forest Vs Neighourest Neighbour(DTW)(3)

19) All Models