Python 中的循环逻辑回归

Question

我想在数据框上构建一个 for 循环，目的是创建一个包含每只股票准确度得分的 df。

一只股票的模型工作正常，但 for 循环不工作 anything.Below 是 df 的输出，这不是完整的 df。

Date        Close   ticker  rating  price   returns direction   long direction 
2021-02-06  21.8   AD.AS    1       21.8    -0.02   -1          1
2021-02-06  21.8   AD.AS    1       21.8    -0.02   -1          1
2021-02-06  21.8   APPL     1       153     -0.02   -1          1
2021-02-06  21.8   APPL     1       153     -0.02   -1          1

stock_df['ticker'].unique()
array(['CSCO', 'IBM', 'AMZN', 'AD.AS'], dtype=object)

我们终于想出了如何做到这一点，请参阅下面我们现在使用的代码。

#for loop test
#Split data into training and test sets
stock_df = stock_df.dropna()

preds_sell = {}
y_test={}
temp = {}

# loop on every type
for ticker in stock_df['ticker'].unique():
    # slice
    stock_slice = stock_df[stock_df['ticker'] == ticker]

    X = stock_slice.drop(['long direction', 'BuyFlag','SellFlag', 'Date', 'ticker'], axis=1)
    y = stock_slice['SellFlag']

    X_train, X_test, y_train, y_test = sklearn.model_selection.train_test_split(X, y, test_size = 0.25, random_state = 5, shuffle=False)

    #Creatig duplicate for back testing
    X_test_2 = X_test

    #building logistic regression model on training data
    model1= LogisticRegression(random_state=0, multi_class='ovr', penalty='none', solver='newton-cg', class_weight={0:0.6, 1:0.4}).fit(X_train, y_train)
    preds_sell[ticker] = model1.predict(X_test)

    #Accuracy statistics
    print('Accuracy Score:', metrics.accuracy_score(y_test, preds_sell[ticker]))  

    #Create classification report
    class_report=classification_report(y_test, preds_sell[ticker])
    print(class_report)

    # build dataframe with all your results
    temp[ticker] = metrics.accuracy_score(y_test, preds_sell[ticker])

temp = pd.DataFrame.from_dict(temp, orient ='index').reset_index()
temp = final_df.rename(columns={'index': 'ticker', 0: 'Accuracy Score'})
final_df = final_df.append(temp)

Answer 1

您没有更新 for-loop 中名为 result 的变量值。

Answer 2

#for loop test
#Split data into training and test sets
stock_df = stock_df.dropna()

preds_sell = {}
y_test={}
temp = {}

# loop on every type
for ticker in stock_df['ticker'].unique():
    # slice
    stock_slice = stock_df[stock_df['ticker'] == ticker]

    X = stock_slice.drop(['long direction', 'BuyFlag','SellFlag', 'Date', 'ticker'], axis=1)
    y = stock_slice['SellFlag']

    X_train, X_test, y_train, y_test = sklearn.model_selection.train_test_split(X, y, test_size = 0.25, random_state = 5, shuffle=False)

    #Creatig duplicate for back testing
    X_test_2 = X_test

    #building logistic regression model on training data
    model1= LogisticRegression(random_state=0, multi_class='ovr', penalty='none', solver='newton-cg', class_weight={0:0.6, 1:0.4}).fit(X_train, y_train)
    preds_sell[ticker] = model1.predict(X_test)

    #Accuracy statistics
    print('Accuracy Score:', metrics.accuracy_score(y_test, preds_sell[ticker]))  

    #Create classification report
    class_report=classification_report(y_test, preds_sell[ticker])
    print(class_report)

    # build dataframe with all your results
    temp[ticker] = metrics.accuracy_score(y_test, preds_sell[ticker])

temp = pd.DataFrame.from_dict(temp, orient ='index').reset_index()
temp = final_df.rename(columns={'index': 'ticker', 0: 'Accuracy Score'})
final_df = final_df.append(temp)

Python 中的循环逻辑回归

For loop Logistic regression in Python

python

for-loop

machine-learning

pandas

logistic-regression