230603

linear regression/1.md

@@ -0,0 +1,18 @@
+# 1.py
+
+$$
+y = wx + b
+$$
+
+## 梯度下降算法
+
+$$
+b_gradient += -\frac{2}{N} \left(y - (w_current \cdot x + b_current)\right)
+$$
+
+
+
+$$
+w_gradient += -\frac{2}{N} \cdot x \cdot \left(y - (w_current \cdot x + b_current)\right)
+$$
+

linear regression/1.py

@@ -19,8 +19,8 @@ def step_gradient(b_current, w_current, points, learningRate):
     for i in range(len(points)):
         x = points[i][0]
         y = points[i][1]
-        b_gradient += -(2 / N) * (y - (w_current * x + b_current) + b_current)
+        b_gradient += -(2 / N) * (y - (w_current * x + b_current))
-        w_gradient += -(2 / N) * x * (y - (w_current * x + b_current + b_current))
+        w_gradient += -(2 / N) * x * (y - (w_current * x + b_current))
     new_b = b_current - (learningRate * b_gradient)
     new_w = w_current - (learningRate * w_gradient)
     return [new_b, new_w]

linear regression/1_cuda.py

@@ -0,0 +1,67 @@
+import matplotlib.pyplot as plt
+import numpy as np
+import torch
+
+# 线性回归训练代码
+def compute_error_for_line_given_points(b, w, points):
+    totalError = 0
+    N = float(len(points))
+    for i in range(len(points)):
+        x = points[i][0]
+        y = points[i][1]
+        totalError += (y - (w * x + b)) ** 2
+    return totalError / N
+
+def step_gradient(b_current, w_current, points, learningRate):
+    b_gradient = torch.tensor(0.0, device=points.device)
+    w_gradient = torch.tensor(0.0, device=points.device)
+    N = float(len(points))
+    for i in range(len(points)):
+        x = points[i][0]
+        y = points[i][1]
+        b_gradient += -(2 / N) * (y - (w_current * x + b_current))
+        w_gradient += -(2 / N) * x * (y - (w_current * x + b_current))
+    new_b = b_current - (learningRate * b_gradient)
+    new_w = w_current - (learningRate * w_gradient)
+    return [new_b, new_w]
+
+def gradient_descent_runner(points, starting_b, starting_w, learningRate, num_iterations):
+    b = torch.tensor(starting_b, device=points.device)
+    w = torch.tensor(starting_w, device=points.device)
+    for i in range(num_iterations):
+        b, w = step_gradient(b, w, points, learningRate)
+    return [b, w]
+
+def run():
+    points_np = np.genfromtxt("data1.csv", delimiter=',').astype(np.float32)
+    points = torch.tensor(points_np, device='cuda')
+    learning_rate = 0.0001
+    initial_b = 0.0
+    initial_w = 0.0
+    num_iterations = 100000
+    [b, w] = gradient_descent_runner(points, initial_b, initial_w, learning_rate, num_iterations)
+    print("After gradient descent at b={0}, w={1}, error={2}".format(b.item(), w.item(),
+                                                                   compute_error_for_line_given_points(b, w, points)))
+    return b.item(), w.item()
+
+# 运行线性回归
+final_b, final_w = run()
+
+# 绘制图像
+points_np = np.genfromtxt("data1.csv", delimiter=',').astype(np.float32)
+x = points_np[:, 0]
+y = points_np[:, 1]
+
+x_range = np.linspace(min(x), max(x), 100)
+y_pred = final_w * x_range + final_b
+
+plt.figure(figsize=(8, 6))
+plt.scatter(x, y, color='blue', label='Original data')
+plt.plot(x_range, y_pred, color='red', label='Fitted line')
+plt.xlabel('X')
+plt.ylabel('Y')
+plt.title('Fitting a line to random data')
+plt.legend()
+plt.grid(True)
+plt.savefig('print1.png')
+plt.show()

linear regression/1x.py

@@ -0,0 +1,72 @@
+import matplotlib.pyplot as plt
+import numpy as np
+import torch
+
+
+# 线性回归训练代码
+def compute_error_for_line_given_points(b, w, points):
+    totalError = 0
+    N = float(len(points))
+    for i in range(len(points)):
+        x = points[i][0]
+        y = points[i][1]
+        totalError += (y - (w * x + b)) ** 2
+    return totalError / N
+
+
+def step_gradient(b_current, w_current, points, learningRate):
+    b_gradient = torch.tensor(0.0, device=points.device, dtype=torch.float32)
+    w_gradient = torch.tensor(0.0, device=points.device, dtype=torch.float32)
+    N = float(len(points))
+    for i in range(len(points)):
+        x = points[i][0]
+        y = points[i][1]
+        b_gradient += -(2 / N) * (y - (w_current * x + b_current))
+        w_gradient += -(2 / N) * x * (y - (w_current * x + b_current))
+    new_b = b_current - (learningRate * b_gradient)
+    new_w = w_current - (learningRate * w_gradient)
+    return [new_b, new_w]
+
+
+def gradient_descent_runner(points, starting_b, starting_w, learningRate, num_iterations):
+    b = torch.tensor(starting_b, device=points.device, dtype=torch.float32)
+    w = torch.tensor(starting_w, device=points.device, dtype=torch.float32)
+    for i in range(num_iterations):
+        b, w = step_gradient(b, w, points, learningRate)
+    return [b, w]
+
+
+def run():
+    points_np = np.genfromtxt("data1.csv", delimiter=',').astype(np.float32)
+    points = torch.tensor(points_np, device='mps')
+    learning_rate = 0.0001
+    initial_b = 0.0
+    initial_w = 0.0
+    num_iterations = 100000
+    [b, w] = gradient_descent_runner(points, initial_b, initial_w, learning_rate, num_iterations)
+    print("After gradient descent at b={0},w={1},error={2}".format(b.item(), w.item(),
+                                                                   compute_error_for_line_given_points(b, w, points)))
+    return b.item(), w.item()
+
+
+# 运行线性回归
+final_b, final_w = run()
+
+# 绘制图像
+points_np = np.genfromtxt("data1.csv", delimiter=',').astype(np.float32)
+x = points_np[:, 0]
+y = points_np[:, 1]
+
+x_range = np.linspace(min(x), max(x), 100)
+y_pred = final_w * x_range + final_b
+
+plt.figure(figsize=(8, 6))
+plt.scatter(x, y, color='blue', label='Original data')
+plt.plot(x_range, y_pred, color='red', label='Fitted line')
+plt.xlabel('X')
+plt.ylabel('Y')
+plt.title('Fitting a line to random data')
+plt.legend()
+plt.grid(True)
+plt.savefig('print1.png')
+plt.show()

linear regression/data1.csv

@@ -1,100 +1,100 @@
-0.0,2.4360562173289115
+0.0,3.267264598063918
-0.10101010101010101,2.4288710820592065
+0.10101010101010101,3.3715980311448757
-0.20202020202020202,3.677943201375977
+0.20202020202020202,3.624529195538264
-0.30303030303030304,3.5029234515863217
+0.30303030303030304,2.2426026435865785
-0.40404040404040403,4.007715980839878
+0.40404040404040403,2.881354128303834
-0.5050505050505051,3.95999321461469
+0.5050505050505051,4.108915299601898
-0.6060606060606061,3.220853916066527
+0.6060606060606061,3.2833072841616344
-0.7070707070707071,3.2211460206798623
+0.7070707070707071,3.401314666490608
-0.8080808080808081,4.2516957270374505
+0.8080808080808081,3.4471224977820083
-0.9090909090909091,4.311826715084292
+0.9090909090909091,4.597483332850038
-1.0101010101010102,4.153966583608258
+1.0101010101010102,4.1948230194917615
-1.1111111111111112,4.224290328721461
+1.1111111111111112,4.770110614428998
-1.2121212121212122,4.551324602105953
+1.2121212121212122,4.3466984672473545
-1.3131313131313131,5.157200101408408
+1.3131313131313131,4.085374736788284
-1.4141414141414141,5.199011258508288
+1.4141414141414141,4.860667770156386
-1.5151515151515151,5.248911218901843
+1.5151515151515151,5.367460741298345
-1.6161616161616161,5.789628423512512
+1.6161616161616161,5.1076464505556585
-1.7171717171717171,5.126592322934872
+1.7171717171717171,4.517380143483942
-1.8181818181818181,4.546631494636344
+1.8181818181818181,6.028333717306668
-1.9191919191919191,5.7260434379514065
+1.9191919191919191,5.268642728341781
-2.0202020202020203,5.607446671816119
+2.0202020202020203,5.2032646463511885
-2.121212121212121,5.401744626671172
+2.121212121212121,5.776924577040542
-2.2222222222222223,5.568078510495838
+2.2222222222222223,5.914239664440679
-2.323232323232323,6.136817713051054
+2.323232323232323,6.195294604152318
-2.4242424242424243,5.399802896696589
+2.4242424242424243,6.67461745554651
-2.525252525252525,6.7465591899811415
+2.525252525252525,6.62895898059055
-2.6262626262626263,6.510002771256968
+2.6262626262626263,6.423496434474387
-2.727272727272727,6.194107987238278
+2.727272727272727,6.520626001853953
-2.8282828282828283,6.280445605022811
+2.8282828282828283,6.252851138402289
-2.929292929292929,6.413289184504817
+2.929292929292929,7.045662416151556
-3.0303030303030303,8.178951965980268
+3.0303030303030303,7.062687254815803
-3.131313131313131,7.438933818741419
+3.131313131313131,6.950015155958233
-3.2323232323232323,8.161193108124682
+3.2323232323232323,7.71420449451215
-3.3333333333333335,6.466487953447159
+3.3333333333333335,7.536987534120887
-3.4343434343434343,7.6815673373443385
+3.4343434343434343,8.408446293914908
-3.5353535353535355,7.412509123916619
+3.5353535353535355,8.281116903817127
-3.6363636363636362,7.712231039046388
+3.6363636363636362,6.862064335470844
-3.7373737373737375,7.512155302443977
+3.7373737373737375,8.455114086555362
-3.8383838383838382,8.169468174953455
+3.8383838383838382,8.610569256326439
-3.9393939393939394,8.201406255891817
+3.9393939393939394,8.695172603505283
-4.040404040404041,9.413915839209679
+4.040404040404041,7.987174933011048
-4.141414141414141,7.2131607261403
+4.141414141414141,8.484042583282307
-4.242424242424242,8.244196707034996
+4.242424242424242,8.152218590549857
-4.343434343434343,8.059400613529792
+4.343434343434343,8.810112829362456
-4.444444444444445,9.127093042087843
+4.444444444444445,9.098520210970904
-4.545454545454545,8.232456814994503
+4.545454545454545,9.315991463976044
-4.646464646464646,9.026988954051767
+4.646464646464646,9.266562387635002
-4.747474747474747,8.936405824368308
+4.747474747474747,8.457655714255173
-4.848484848484849,8.838334259675397
+4.848484848484849,8.577190426286784
-4.94949494949495,9.717080564295035
+4.94949494949495,9.992687218959654
-5.05050505050505,9.635892324495916
+5.05050505050505,9.949888900251127
-5.151515151515151,10.802758752616178
+5.151515151515151,10.112370557219064
-5.252525252525253,9.889268431487773
+5.252525252525253,10.250084050804231
-5.353535353535354,9.262021983987134
+5.353535353535354,9.675169646286898
-5.454545454545454,9.905732041295009
+5.454545454545454,9.790565255890696
-5.555555555555555,9.697006564677089
+5.555555555555555,9.91666488079517
-5.656565656565657,10.435437946557755
+5.656565656565657,10.325538746448835
-5.757575757575758,10.257651825530608
+5.757575757575758,9.77548528051785
-5.858585858585858,11.394734709569004
+5.858585858585858,10.55371401462777
-5.959595959595959,10.872621683473387
+5.959595959595959,10.757696722894282
-6.0606060606060606,10.750944058491058
+6.0606060606060606,10.893354131765726
-6.161616161616162,11.375400587831757
+6.161616161616162,12.049342074375708
-6.262626262626262,11.834436555701465
+6.262626262626262,10.936118426966079
-6.363636363636363,11.536088544119654
+6.363636363636363,11.031578580287063
-6.4646464646464645,11.261555999325722
+6.4646464646464645,11.713471927909302
-6.565656565656566,12.529961808490153
+6.565656565656566,12.343664117608101
-6.666666666666667,12.19345219105891
+6.666666666666667,12.067856620638729
-6.767676767676767,11.950653180245155
+6.767676767676767,11.814430199711552
-6.8686868686868685,12.176773142948385
+6.8686868686868685,11.123516182999314
-6.96969696969697,12.055083206520518
+6.96969696969697,12.496962644202316
-7.070707070707071,13.498633194384489
+7.070707070707071,12.767487737755147
-7.171717171717171,12.542518727882712
+7.171717171717171,12.632934104476211
-7.2727272727272725,13.318372269865769
+7.2727272727272725,12.728225932468364
-7.373737373737374,12.542630883166513
+7.373737373737374,12.97630338885533
-7.474747474747475,12.93490122675753
+7.474747474747475,12.896220727223701
-7.575757575757575,14.4040220344926
+7.575757575757575,13.047808359849581
-7.6767676767676765,13.314367294113964
+7.6767676767676765,13.110443597152527
-7.777777777777778,14.061236496574551
+7.777777777777778,12.921358752181128
-7.878787878787879,12.686346979737731
+7.878787878787879,13.30038615173782
-7.979797979797979,14.024375221983842
+7.979797979797979,13.836945395153705
-8.080808080808081,13.7042096336008
+8.080808080808081,13.054484897082014
-8.181818181818182,13.342730021126272
+8.181818181818182,14.01038452336861
-8.282828282828282,14.136548357864573
+8.282828282828282,13.643336312636018
-8.383838383838384,14.619569834949138
+8.383838383838384,14.564671365817466
-8.484848484848484,14.01453898823226
+8.484848484848484,14.040540515755758
-8.585858585858587,15.154877807203663
+8.585858585858587,14.57992522742261
-8.686868686868687,14.081910297898048
+8.686868686868687,14.88631275019171
-8.787878787878787,14.474564310016353
+8.787878787878787,14.021963220008606
-8.88888888888889,14.966525346412723
+8.88888888888889,15.068155050128949
-8.98989898989899,15.526107019435932
+8.98989898989899,15.083538874549268
-9.09090909090909,14.352357736719853
+9.09090909090909,15.417748308319911
-9.191919191919192,15.843742065895144
+9.191919191919192,14.89714205401168
-9.292929292929292,15.787083172159111
+9.292929292929292,14.534676091762206
-9.393939393939394,15.211828607109144
+9.393939393939394,15.556467883324295
-9.494949494949495,15.845176532492374
+9.494949494949495,15.525938847099864
-9.595959595959595,15.622518083688107
+9.595959595959595,15.560767751324764
-9.696969696969697,15.589081237426006
+9.696969696969697,15.982790914773943
-9.797979797979798,15.511248085690712
+9.797979797979798,16.062079721169738
-9.8989898989899,16.27050774059862
+9.8989898989899,16.232818049890696
-10.0,16.1105549896166
+10.0,17.053472736980353

linear regression/plt_print.py

@@ -3,12 +3,13 @@ import matplotlib.pyplot as plt
 
 # 原始数据
 points = np.genfromtxt("data1.csv", delimiter=',')
+
 x = points[:, 0]
 y = points[:, 1]
 
 # 拟合直线
 x_range = np.linspace(min(x), max(x), 100)
-y_pred = 1.6455038785934448 * x_range + 1.827562689781189
+y_pred = 0.3880246877670288 * x_range + 1.7214288711547852
 
 # 绘图
 plt.figure(figsize=(8, 6))