Softmax Regression

Softmax Regression python实现。

权重训练文件：

import numpy as np
import time

# file_path1 = open("./Data.txt", "r")
file_path1 = 'Data.txt'


def load_dataset(fileinput):
    '''
    导入数据
    input: fileinput(string)数据文件的位置
    output: feature_data(mat)特征样例，一行一个训练样例
            label_data(mat)标签数据，列向量
            k(int)类别数
    '''
    f = open(fileinput, "r")
    feature_data = []
    label_data = []
    for line in f.readlines():
        feature_temp = []
        feature_temp.append(1)
        lines = line.strip().split()

        for i in range(len(lines)-1):
            feature_temp.append(float(lines[i]))
        label_data.append(float(lines[-1]))

        feature_data.append(feature_temp)
    f.close()
    return np.mat(feature_data), np.mat(label_data).T, len(set(label_data))

def cost(err, label_data):
    '''计算损失函数值
    input:  err(mat):exp的值
            label_data(mat):标签的值
    output: sum_cost / m(float):损失函数的值
    '''
    m = np.shape(err)[0]
    sum_cost = 0.0
    for i in range(m):
        if err[i, int(label_data[i, 0])] / np.sum(err[i, :]) > 0:
            sum_cost -= np.log(err[i, label_data[i, 0]] / np.sum(err[i, :]))
        else:
            sum_cost -= 0
    return sum_cost / m


def gradientAscent(feature_data, label_data, k, maxCycle, alpha):
    '''利用梯度下降法训练Softmax模型
    input:  feature_data(mat):特征
            label_data(mat):标签
            k(int):类别的个数
            maxCycle(int):最大的迭代次数
            alpha(float):学习率
    output: weights(mat)：权重,每列对应一个类别
    '''
    m, n = np.shape(feature_data)
    weights = np.mat(np.ones((n, k)))
    i = 0
    while i <= maxCycle:
        error = np.exp(feature_data * weights)

        rowsum = -error.sum(axis=1)
        rowsum = rowsum.repeat(k, axis=1)
        error = error / rowsum
        # print(m)
        for x in range(m):
            error[x, int(label_data[x, 0])-1] = error[x, int(label_data[x, 0])-1] + 1
            # print("hahaha")
        weights = weights + (alpha / m) * feature_data.T * error
        i += 1
    return weights


def save_model(file_name, weights):

    '''保存最终的模型
    input:  file_name(string):保存的文件名
            weights(mat):softmax模型
    '''
    f_w = open(file_name, "w")
    m, n = np.shape(weights)
    for i in range(m):
        w_tmp = []
        for j in range(n):
            w_tmp.append(str(weights[i, j]))
        f_w.write("\t".join(w_tmp) + "\n")
    f_w.close()


def test():
    x, y, z = load_dataset(file_path1)

    print('开始训练模型权重', time.strftime("%a %b %d %H:%M:%S %Y", time.localtime()))
    weights = gradientAscent(x, y, z, 1000, 0.8)
    print('结束训练模型权重', time.strftime("%a %b %d %H:%M:%S %Y", time.localtime()))
    print('保存模型权重文件:weights')
    save_model("weights", weights)


if __name__ == "__main__":

    test()

测试训练权重文件：

'''

'''
import numpy as np
import random as rd


def load_weights(weights_path):
    '''导入训练好的Softmax模型
    input:  weights_path(string)权重的存储位置
    output: weights(mat)将权重存到矩阵中
            m(int)权重的行数
            n(int)权重的列数
    '''
    f = open(weights_path)
    w = []
    for line in f.readlines():
        w_tmp = []
        lines = line.strip().split("\t")
        for x in lines:
            w_tmp.append(float(x))
        w.append(w_tmp)
    f.close()
    weights = np.mat(w)
    m, n = np.shape(weights)
    return weights, m, n


def load_dataset(fileinput):
    '''
    导入数据
    input: fileinput(string)数据文件的位置
    output: feature_data(mat)特征样例，一行一个训练样例
            label_data(mat)标签数据，列向量
            k(int)类别数
    '''
    f = open(fileinput)
    feature_data = []
    label_data = []
    for line in f.readlines():
        feature_temp = []
        feature_temp.append(1)
        lines = line.strip().split()

        for i in range(len(lines) - 1):
            feature_temp.append(float(lines[i]))
        label_data.append(float(lines[-1]))

        feature_data.append(feature_temp)
    f.close()
    return np.mat(feature_data), np.mat(label_data).T, len(set(label_data))


def load_data(num, m):
    '''导入测试数据
    input:  num(int)生成的测试样本的个数
            m(int)样本的维数
    output: testDataSet(mat)生成测试样本
    '''
    testDataSet = np.mat(np.ones((num, m)))
    print(testDataSet)
    for i in range(num):
        testDataSet[i, 1] = rd.random() * 6 - 3  # 随机生成[-3,3]之间的随机数
        testDataSet[i, 2] = rd.random() * 15  # 随机生成[0,15]之间是的随机数
        print(testDataSet)
    return testDataSet


def predict(test_data, weights):
    '''利用训练好的Softmax模型对测试数据进行预测
    input:  test_data(mat)测试数据的特征
            weights(mat)模型的权重
    output: h.argmax(axis=1)所属的类别
    '''
    h = test_data * weights
    return h.argmax(axis=1)  # 获得所属的类别


def save_result(file_name, result):
    '''保存最终的预测结果
    input:  file_name(string):保存最终结果的文件名
            result(mat):最终的预测结果
    '''
    f_result = open(file_name, "w")
    m = np.shape(result)[0]
    for i in range(m):
        f_result.write(str(result[i, 0]) + "\n")
    f_result.close()


def findamenum(num1, num2):
    i = j = 0
    for lc in range(len(num1)):
        if num1[lc] == num2[lc]:
            i += 1
        else:
            j += 1
    return i / len(num1)


if __name__ == "__main__":
    # 1、导入Softmax模型
    print("---------- 1.load model ------------")
    w, m, n = load_weights("weights")
    # 2、导入测试数据
    print("---------- 2.load data ------------")
    file_path1 = 'test_dataset.txt'
    x, y, z = load_dataset(file_path1)

    # test_data = load_data(4000, m)
    # 3、利用训练好的Softmax模型对测试数据进行预测
    print("---------- 3.get Prediction ------------")
    result = predict(x, w) + 1

    # 4、保存最终的预测结果
    print("---------- 4.save prediction ------------")
    save_result("result", result)
    # 测试数据79个
    #
    print("预测概率")
    print(findamenum(y, result))

数据集：训练集，测试集。

数据集分布文件：

import numpy as np
# import matplotlib.pyplot as plt
# from matplotlib.font_manager import FontProperties

a = np.loadtxt('Data.txt')

b = a[:, 0: 2]

print(a)


def gen_dataset():
    # from sklearn.datasets import load_iris
    # from sklearn.model_selection import train_test_split
    # from sklearn.datasets import make_blobs
    import matplotlib.pyplot as pltt
    # 绘制数据分布
    pltt.figure(figsize=(6, 4))
    # 显示图大小
    pltt.scatter(a[:, 0], a[:, 1], c=a[:, 2])
    # 获得数据特征和类标签
    pltt.title("20208223056")
    pltt.xlabel("x feature")
    pltt.ylabel("y feature")
    pltt.show()
######


if __name__ == "__main__":

    gen_dataset()

下载  提取码：wlo3