源码：

由于刚开始写，我只写了一部分。

简易KNN

python">def GetKNNSoreByN(X, y, n_neighbors):
    """
    :param X: data 特征值
    :param y: aim 目标值
    :param n_neighbors K值
    :return: score 预测结果
    """
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
    transform = StandardScaler()
    X_train = transform.fit_transform(X_train)
    X_test = transform.fit_transform(X_test)
    clf = KNeighborsClassifier(n_neighbors=n_neighbors)
    clf.fit(X_train, y_train)
    y_pre = clf.predict(X_test)
    return sum(y_pre == y_test) / y_pre.shape[0]

解释：这里数据标准化使用的是StandardScaler、对传进来的数据集进行了切割，实际上数据足够多的话没必要进行切割，但这里考虑到数据较少的情况进行了切割。

网格搜索版KNN

python">def GetKNNScoreByGridSearchCV(X, y, param_grid: dict={'n_neighbors': [i for i in range(1,10,1)]}):
    """
    :param X:data 特征值
    :param y:aim 目标值
    :param param_grid: GridSearchCV param 传递给GridSearchCV的参数
    :return: best params and best score for KNN最好的参数表以及最佳准确率
    """
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
    transform = StandardScaler()
    X_train = transform.fit_transform(X_train)
    X_test = transform.fit_transform(X_test)
    estimator = KNeighborsClassifier()
    estimator = GridSearchCV(estimator, param_grid=param_grid, cv=5, verbose=0)
    estimator.fit(X_train, y_train)

    return estimator.best_params_, estimator.best_score_

说明：默认的CV为5，verbose=0，这个是经验判断，若有好的建议，可以一起讨论
补充：一般情况K值不超过10，所以默认值给的1到10

正则版线性回归

python">def linear_model_regular(data):
    """
    regular
    :param data: data 数据
    :return: coef 系数列表, intercept 截距, error 均方误差
    """
    x_train, x_test, y_train, y_test = train_test_split(data.data, data.target, random_state=0)
    transfer = StandardScaler()
    x_train = transfer.fit_transform(x_train)
    x_test = transfer.transform(x_test)
    estimator = LinearRegression()
    estimator.fit(x_train, y_train)
    y_predict = estimator.predict(x_test)
    error = mean_squared_error(y_test, y_predict)  # 均方误差

    return estimator.coef_, estimator.intercept_, error

梯度下降版线性回归：

python">def linear_model_gradient(data):
	"""
	gradient descent
    :param data: data 数据
    :return: coef 系数列表, intercept 截距, error 均方误差
    """
    x_train, x_test, y_train, y_test = train_test_split(data.data, data.target, random_state=0)
    transfer = StandardScaler()
    x_train = transfer.fit_transform(x_train)
    x_test = transfer.transform(x_test)
    estimator = SGDRegressor(max_iter=1000)
    estimator.fit(x_train, y_train)
    y_predict = estimator.predict(x_test)
    error = mean_squared_error(y_test, y_predict)  # 均方误差

    return estimator.coef_, estimator.intercept_, error

防止重复库

python">def setup_module(module):
    """
	Prevent multiple uses of the same library
	"""
    # Check if a random seed exists in the environment, if not create one.
    _random_seed = os.environ.get('SKLEARN_SEED', None)
    if _random_seed is None:
        _random_seed = np.random.uniform() * np.iinfo(np.int32).max
    _random_seed = int(_random_seed)
    print("I: Seeding RNGs with %r" % _random_seed)
    np.random.seed(_random_seed)
    random.seed(_random_seed)

安装方法以及更新命令

安装

python">pip install techlearn

更新

python">pip3 install --upgrade techlearn

测试代码以及截图（部分）

KNN&鸢尾花

python">X, y = datasets.load_iris(return_X_y=True)
param_grid = {'n_neighbors': [1, 3, 5, 7]}
print(GetKNNSoreByN(X, y, 3))
print(GetKNNScoreByGridSearchCV(X, y, param_grid=param_grid))
print(GetKNNScoreByGridSearchCV(X, y))

线性回归&波士顿房价预测

python">    data = load_boston()
    coef_, intercept_, error = linear_model_gradient(data=data)
    print("模型中的系数为:\n", coef_)
    print("模型中的偏置为:\n", intercept_)
    print("误差为:\n", error)