python - 拟合for循环后如何保存sklearn模型

Question

我正在使用管道将两个 sklearn 模型放入 for 循环中，并尝试将它们单独命名并稍后使用此管道进行预测，

names = ["Linear", "Elastic"]
models = [LinearRegression(),ElasticNet()]
model_fit = [x + 'fit' for x in names]

for name, model, fits in zip(names, models, model_fit):
    steps = [('scaler', MinMaxScaler()),(name, model)]
    pipe = Pipeline(steps)
    fits = pipe.fit(x_train, y_train)

但我无法在不同的循环中预测它。

for i, model in enumerate(model_fit):
   predictions = model.predict(x_test)

收到以下错误

'str' object has no attribute 'predict'

如何以不同的名称拟合和检索这两个模型？

Answer 1

您的 model_fit 变量是一个字符串数组，因此您不应在第二个 for 循环中使用它。您应该做的是将模型存储在另一个数组中并使用它。

names = ["Linear", "Elastic"]
models = [LinearRegression(),ElasticNet()]
model_fit = [x + 'fit' for x in names]
fitted_models = []

for name, model, fits in zip(names, models, model_fit):
    steps = [('scaler', MinMaxScaler()),(name, model)]
    pipe = Pipeline(steps)
    fits = pipe.fit(x_train, y_train)
    fitted_models.append(fits)

for i, model in enumerate(fitted_models):
   predictions = model.predict(x_test)