titanic.py 文件源码

def normalize_data():
    data = []
    with open(folder + filename, 'rb') as csvfile:
        spamreader = csv.DictReader(csvfile)
        for row in spamreader:
            for key in keys_to_remove:
                del row[key]

            row['Sex'] = gender_to_number[row['Sex']]
            row['Embarked'] = port_to_number[row['Embarked']]
            row['Age'] = 0 if row['Age'] == "" else float(row['Age'])
            row['Parch'] = 0 if row['Parch'] == "" else int(row['Parch'])
            row['Pclass'] = 3 if row['Pclass'] == "" else int(row['Pclass'])
            row['Survived'] = int(row['Survived'])
            row['SibSp'] = 0 if row['SibSp'] == "" else int(row['SibSp'])
            row['Cabin'] = 0 if row['Cabin'] == "" else 1

            data.append(row)

    return data