learn.txt

@Author: Atul Sahay <atul>
@Date:   2018-08-13T11:52:05+05:30
@Email:  atulsahay01@gmail.com
@Filename: learn.txt
@Last modified by:   atul
@Last modified time: 2018-08-19T20:34:36+05:30


for cost in 500 ranges please multiply it with 2


power = 7

cost = 514.897004562180541143

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,1.09)

2.

cost : 514.202864048305286815

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,1.05)

3.

cost : 513.298057622226679086

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,1)

4.

error : 33.11011
cost : 509.208405682665670611

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.9)

5.

cost : 508.051560120712451862

learn = 0.1
iterations = 1000

# weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.75)

6.

error = 33.11011

cost : 509.208405682665670611

learn = 0.1
iterations = 1000

# weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.8)


power  = 5

1.

error 33.19
cost = 502.623970685411734394
learn = 0.1
iterations = 1000

# weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.8)


power = 9

1.

error = 33.0666
cost = 512.71414308990935313
learn = 0.1

# weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.8)

2.

error = 32.949494
cost = 510.475963283599583065
learn = 0.1

weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.7)

3.

error = 32.933313

cost = 508
cost = 510.475963283599583065
learn = 0.1

weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.6)

4.

error = 32.91831
cost = 498
learn = 0.1

weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.3)

5.

error = 491.07181931555783192

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.1)

6

error = 32.99

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.1)


power = 11

1.

error 32.8777

cost 501.6644420796

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.3)


2.

teration 999 | Cost: 503.720206335113402929 Cost_reg 516.5807
31023826956516

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.3)

power = 17

1.

error = 32.811

learn = 0.1
  iterations = 1000

  # weights = update_weights(learn,iterations,x_train.values,y_train.values,weights)
  weights = gradientDescent(x_train.values, y_train.values, weights, 0.009, 32000,iterations,0.2)




############################ NEW ################################################
1. submitted -------- 32.76269
iteration 730 | Cost_reg 1036.102647140478893562
Cost_valid_old 1122.035566177721875647
Cost_valid_curr 1122.035576808738369436
lambda: 10000.000000    learn:0.001000

Diff:85.921657735369080910

2.

teration 715 | Cost_reg 1033.482797809866497118
Cost_valid_old 1115.387826085673623311
Cost_valid_curr 1115.387854471180162363
lambda: 8000.000000     learn:0.001000

Diff:81.891807065056127612

3.
teration 822 | Cost_reg 1026.430208265527426192
Cost_valid_old 1101.262472840467808055
Cost_valid_curr 1101.262488131178770345
lambda: 5000.000000     learn:0.001000

Diff:74.817013307808565514

4.
teration 1184 | Cost_reg 1015.217789628799891943
Cost_valid_old 1086.290447758664413413
Cost_valid_curr 1086.290452804756341720
lambda: 3000.000000     learn:0.001000

Diff:71.057872969656500572

5.
teration 1688 | Cost_rg 1003.318374605609619721
Cost_valid_old 1073.854306582763456390
Cost_valid_curr 1073.854308007453255414
lambda: 2000.000000     learn:0.001000

Diff:70.522778926420073731

6.

teration 1616 | Cost_reg 1004.883908228200425583
Cost_valid_old 1075.407808527469569526
Cost_valid_curr 1075.407813744125860467
lambda: 2100.000000     learn:0.001000

Diff:70.510511110830407233

7.

teration 1550 | Cost_reg 1006.349123106250090132
Cost_valid_old 1076.872010202909450527
Cost_valid_curr 1076.872013795853490592
lambda: 2200.000000     learn:0.001000

Diff:70.509278114476728661

8. submit

teration 748 | Cost_reg 1037.145539877368491943
Cost_valid_old 1124.749272929149128686
Cost_valid_curr 1124.749298215575436188
lambda: 11000.000000    learn:0.001000

Diff:87.593491049141221083

9. submit

teration 511 | Cost_reg 1037.261755384825164583
Cost_valid_old 1119.580632949880282467
Cost_valid_curr 1119.580636112286811112
lambda: 110900.000000   learn:0.001000   P-Norm = 4

Diff:82.300452014113261612

10. submit

teration 367 | Cost_reg 1037.896387546817322800
Cost_valid_old 1114.612586483802260773
Cost_valid_curr 1114.612589134222162102
lambda: 119000.000000   learn:0.001000   P-Norm = 6

Diff:76.693289107487544243

11. biases - sigmoidal submitted --- 32.60321

teration 5328 | Cost_reg 1040.179873456502036788
Cost_valid_old 1093.415007206548352769
Cost_valid_curr 1093.415007397611134365
lambda: 740.000000      learn:0.001000

Diff:53.234030937904435632

12. bias - Gaussian submitted ------- 32.66542

teration 335 | Cost_reg 1038.030676439241005937
Cost_valid_old 1099.393839132729226549
Cost_valid_curr 1099.393920349684549365
lambda: 13000.000000    learn:0.001000

Diff:61.324867345521170137

13. lambda = 0 submitted ---- 39.87926

teration 603502 | Cost_reg 760.90590109215122538
Cost_valid_old 726.684941062964298908
Cost_valid_curr 726.684941062966231584
lambda: 0.000000        learn:0.030000

Diff:34.220961581810570351

14. p = 4 norm submitted ----- 32.73488

teration 518 | Cost_reg 1036.877551579356804723
Cost_valid_old 1118.861163783287338447
Cost_valid_curr 1118.861189012485738203
lambda: 109000.000000   learn:0.001000   P-Norm = 4

Diff:81.965035595715107775

15. pnorm = 6 submitted -------- 32.71304

teration 367 | Cost_reg 1037.896387546817322800
Cost_valid_old 1114.612586483802260773
Cost_valid_curr 1114.612589134222162102
lambda: 119000.000000   learn:0.001000   P-Norm = 6

Diff:76.693289107487544243

16. pnorm = 6 and inverse Sigmoid submitted --------- 32.59869

teration 215 | Cost_reg 1048.123778150038333479
Cost_valid_old 1097.412015738770378448
Cost_valid_curr 1097.412015993267459635
lambda: 706.000000      learn:0.001000   P-Norm = 6

Diff:49.284048753764864159