ai策略报错,新人求帮助

问答交流
标签: #<Tag:0x00007f4cdb46d230>

(qwe9527) #1


我的策略报这个错,features只有这个mean(close_0, 20),我想问这个feature写法对吗,为什么策略报keyerror这个错
下面是策略代码

克隆策略
In [18]:
# 基础参数配置
class conf:
    start_date = '2010-01-01'
    end_date='2017-01-01'
    # split_date 之前的数据用于训练,之后的数据用作效果评估
    split_date = '2015-01-01'
    # D.instruments: https://bigquant.com/docs/data_instruments.html
    instruments = D.instruments(start_date, split_date)

    # 机器学习目标标注函数
    # 如下标注函数等价于 min(max((持有期间的收益 * 100), -20), 20) + 20 (后面的M.fast_auto_labeler会做取整操作)
    # 说明:max/min这里将标注分数限定在区间[-20, 20],+20将分数变为非负数 (StockRanker要求标注分数非负整数)
    label_expr = ['return * 100', 'where(label > {0}, {0}, where(label < -{0}, -{0}, label)) + {0}'.format(20)]
    # 持有天数,用于计算label_expr中的return值(收益)
    hold_days = 5

    # 特征 https://bigquant.com/docs/data_features.html,你可以通过表达式构造任何特征
    features = [
        'mean(close_0, 20)'
    ]

# 给数据做标注:给每一行数据(样本)打分,一般分数越高表示越好
m1 = M.fast_auto_labeler.v8(
    instruments=conf.instruments, start_date=conf.start_date, end_date=conf.split_date,
    label_expr=conf.label_expr, hold_days=conf.hold_days,
    benchmark='000300.SHA', sell_at='open', buy_at='open')
# 计算特征数据
m2 = M.general_feature_extractor.v5(
    instruments=conf.instruments, start_date=conf.start_date, end_date=conf.split_date,
    features=conf.features)
# 计算衍生数据
m22 = M.derived_feature_extractor.v2(input_data=m2.data, features=conf.features)

# 数据预处理:缺失数据处理,数据规范化,T.get_stock_ranker_default_transforms为StockRanker模型做数据预处理
m3 = M.transform.v2(
    data=m22.data, transforms=T.get_stock_ranker_default_transforms()+[('.*', None)],
    drop_null=True, astype='int32', except_columns=['date', 'instrument'],
    clip_lower=0, clip_upper=200000000)
# 合并标注和特征数据
m4 = M.join.v2(data1=m1.data, data2=m3.data, on=['date', 'instrument'], sort=True)

# StockRanker机器学习训练
m5 = M.stock_ranker_train.v3(training_ds=m4.data, features=conf.features)


## 量化回测 https://bigquant.com/docs/module_trade.html
# 回测引擎:准备数据,只执行一次
def prepare(context):
    # context.start_date / end_date,回测的时候,为trader传入参数;在实盘运行的时候,由系统替换为实盘日期
    n1 = M.general_feature_extractor.v5(
        instruments=D.instruments(),
        start_date=context.start_date, end_date=context.end_date,
        model_id=context.options['model_id'])
    n2 = M.transform.v2(
        data=n1.data, transforms=T.get_stock_ranker_default_transforms(),
        drop_null=True, astype='int32', except_columns=['date', 'instrument'],
        clip_lower=0, clip_upper=200000000)
    n3 = M.stock_ranker_predict.v2(model_id=context.options['model_id'], data=n2.data)
    context.instruments = n3.instruments
    context.options['predictions'] = n3.predictions

# 回测引擎:初始化函数,只执行一次
def initialize(context):
    # 加载预测数据
    context.ranker_prediction = context.options['predictions'].read_df()

    # 系统已经设置了默认的交易手续费和滑点,要修改手续费可使用如下函数
    context.set_commission(PerOrder(buy_cost=0.0003, sell_cost=0.0013, min_cost=5))
    # 预测数据,通过options传入进来,使用 read_df 函数,加载到内存 (DataFrame)
    # 设置买入的股票数量,这里买入预测股票列表排名靠前的5只
    stock_count = 5
    # 每只的股票的权重,如下的权重分配会使得靠前的股票分配多一点的资金,[0.339160, 0.213986, 0.169580, ..]
    context.stock_weights = T.norm([1 / math.log(i + 2) for i in range(0, stock_count)])
    # 设置每只股票占用的最大资金比例
    context.max_cash_per_instrument = 0.2

# 回测引擎:每日数据处理函数,每天执行一次
def handle_data(context, data):
    # 按日期过滤得到今日的预测数据
    ranker_prediction = context.ranker_prediction[
        context.ranker_prediction.date == data.current_dt.strftime('%Y-%m-%d')]

    # 1. 资金分配
    # 平均持仓时间是hold_days,每日都将买入股票,每日预期使用 1/hold_days 的资金
    # 实际操作中,会存在一定的买入误差,所以在前hold_days天,等量使用资金;之后,尽量使用剩余资金(这里设置最多用等量的1.5倍)
    is_staging = context.trading_day_index < context.options['hold_days'] # 是否在建仓期间(前 hold_days 天)
    cash_avg = context.portfolio.portfolio_value / context.options['hold_days']
    cash_for_buy = min(context.portfolio.cash, (1 if is_staging else 1.5) * cash_avg)
    cash_for_sell = cash_avg - (context.portfolio.cash - cash_for_buy)
    positions = {e.symbol: p.amount * p.last_sale_price
                 for e, p in context.perf_tracker.position_tracker.positions.items()}

    # 2. 生成卖出订单:hold_days天之后才开始卖出;对持仓的股票,按StockRanker预测的排序末位淘汰
    if not is_staging and cash_for_sell > 0:
        equities = {e.symbol: e for e, p in context.perf_tracker.position_tracker.positions.items()}
        instruments = list(reversed(list(ranker_prediction.instrument[ranker_prediction.instrument.apply(
                lambda x: x in equities and not context.has_unfinished_sell_order(equities[x]))])))
        # print('rank order for sell %s' % instruments)
        for instrument in instruments:
            context.order_target(context.symbol(instrument), 0)
            cash_for_sell -= positions[instrument]
            if cash_for_sell <= 0:
                break

    # 3. 生成买入订单:按StockRanker预测的排序,买入前面的stock_count只股票
    buy_cash_weights = context.stock_weights
    buy_instruments = list(ranker_prediction.instrument[:len(buy_cash_weights)])
    max_cash_per_instrument = context.portfolio.portfolio_value * context.max_cash_per_instrument
    for i, instrument in enumerate(buy_instruments):
        cash = cash_for_buy * buy_cash_weights[i]
        if cash > max_cash_per_instrument - positions.get(instrument, 0):
            # 确保股票持仓量不会超过每次股票最大的占用资金量
            cash = max_cash_per_instrument - positions.get(instrument, 0)
        if cash > 0:
            context.order_value(context.symbol(instrument), cash)


# 调用交易引擎
m6 = M.trade.v2(
    instruments=None,
    start_date=conf.split_date,
    end_date=conf.end_date,
    prepare=prepare,
    initialize=initialize,
    handle_data=handle_data,
    order_price_field_buy='open',       # 表示 开盘 时买入
    order_price_field_sell='close',     # 表示 收盘 前卖出
    capital_base=1000000,               # 初始资金
    benchmark='000300.SHA',             # 比较基准,不影响回测结果
    # 通过 options 参数传递预测数据和参数给回测引擎
    options={'hold_days': conf.hold_days, 'model_id': m5.model_id}
)
[2017-09-02 20:47:17.778565] INFO: bigquant: fast_auto_labeler.v8 开始运行..
[2017-09-02 20:47:17.781595] INFO: bigquant: 命中缓存
[2017-09-02 20:47:17.785542] INFO: bigquant: fast_auto_labeler.v8 运行完成[0.006996s].
[2017-09-02 20:47:17.791954] INFO: bigquant: general_feature_extractor.v5 开始运行..
[2017-09-02 20:47:17.794344] INFO: bigquant: 命中缓存
[2017-09-02 20:47:17.795729] INFO: bigquant: general_feature_extractor.v5 运行完成[0.003752s].
[2017-09-02 20:47:17.803286] INFO: bigquant: derived_feature_extractor.v2 开始运行..
[2017-09-02 20:47:17.805501] INFO: bigquant: 命中缓存
[2017-09-02 20:47:17.806523] INFO: bigquant: derived_feature_extractor.v2 运行完成[0.003248s].
[2017-09-02 20:47:17.814781] INFO: bigquant: transform.v2 开始运行..
[2017-09-02 20:47:17.816568] INFO: bigquant: 命中缓存
[2017-09-02 20:47:17.817328] INFO: bigquant: transform.v2 运行完成[0.002549s].
[2017-09-02 20:47:17.823584] INFO: bigquant: join.v2 开始运行..
[2017-09-02 20:47:17.825405] INFO: bigquant: 命中缓存
[2017-09-02 20:47:17.826336] INFO: bigquant: join.v2 运行完成[0.002746s].
[2017-09-02 20:47:17.834955] INFO: bigquant: stock_ranker_train.v3 开始运行..
[2017-09-02 20:47:17.848335] INFO: stock_ranker_train: da6ee378 training: 2573683 rows
---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-18-6086a2b2bdb2> in <module>()
     41 
     42 # StockRanker机器学习训练
---> 43 m5 = M.stock_ranker_train.v3(training_ds=m4.data, features=conf.features)
     44 
     45 

KeyError: 'mean'
In [7]:
M.derived_feature_extractor
Out[7]:
模块:derived_feature_extractor
可用版本(推荐使用最新版本):v2, v1

(小Q) #2

由于最近升级版本了,因此AI训练和预测部分有不少变动,很抱歉给您带来了不少麻烦。

您策略报错的原因是某些接口使用有些变动,请参考下面实验,可克隆到自己策略研究平台。

克隆策略
In [10]:
# 基础参数配置
class conf:
    start_date = '2010-01-01'
    end_date='2017-01-01'
    # split_date 之前的数据用于训练,之后的数据用作效果评估
    split_date = '2015-01-01'
    # D.instruments: https://bigquant.com/docs/data_instruments.html
    instruments = D.instruments(start_date, split_date)[:300]
    hold_days = 5

    features = ['mean(close_0, 20)']

    # 标注
    label_expr = [
        
    # 计算收益:5日收盘价(作为卖出价格)除以明日开盘价(作为买入价格)
    '(shift(close, -5) / shift(open, -1) - 1) * 100',
        
    # 极值处理:用1%和99%分位的值做clip
    'clip(label, all_quantile(label, 0.01), all_quantile(label, 0.99))',
    # 将分数映射到分类,这里使用20个分类
    'all_wbins(label, 20)',
        
    # 过滤掉一字涨停的情况 (设置label为NaN,在后续处理和训练中会忽略NaN的label)
    'where(shift(high, -1) == shift(low, -1), NaN, label)'
    ]
    
    
    
    
    
# 给数据做标注:给每一行数据(样本)打分,一般分数越高表示越好
m1 = M.advanced_auto_labeler.v1(
    instruments=conf.instruments, start_date=conf.start_date, end_date=conf.split_date,
    label_expr=conf.label_expr,
    benchmark='000300.SHA',cast_label_int=True)
# 计算特征数据
m2 = M.general_feature_extractor.v5(
    instruments=conf.instruments, start_date=conf.start_date, end_date=conf.split_date,
    features=conf.features)
# 计算衍生数据
m22 = M.derived_feature_extractor.v2(input_data=m2.data, features=conf.features)

# 数据预处理:缺失数据处理,数据规范化,T.get_stock_ranker_default_transforms为StockRanker模型做数据预处理
m3 = M.transform.v2(data=m22.data, transforms=None, drop_null=True)
     
# 合并标注和特征数据
m4 = M.join.v2(data1=m1.data, data2=m3.data, on=['date', 'instrument'], sort=True)

# StockRanker机器学习训练
m5 = M.stock_ranker_train.v4(training_ds=m4.data, features=conf.features)


## 量化回测 https://bigquant.com/docs/module_trade.html
# 回测引擎:准备数据,只执行一次
def prepare(context):
    # context.start_date / end_date,回测的时候,为trader传入参数;在实盘运行的时候,由系统替换为实盘日期
    n1 = M.general_feature_extractor.v5(
        instruments=D.instruments(),
        start_date=context.start_date, end_date=context.end_date,
        model_id=context.options['model_id'])
    n2 = M.transform.v2(
        data=n1.data, transforms=T.get_stock_ranker_default_transforms(),
        drop_null=True, astype='int32', except_columns=['date', 'instrument'],
        clip_lower=0, clip_upper=200000000)
    n3 = M.stock_ranker_predict.v2(model_id=context.options['model_id'], data=n2.data)
    context.instruments = n3.instruments
    context.options['predictions'] = n3.predictions

# 回测引擎:初始化函数,只执行一次
def initialize(context):
    # 加载预测数据
    context.ranker_prediction = context.options['predictions'].read_df()

    # 系统已经设置了默认的交易手续费和滑点,要修改手续费可使用如下函数
    context.set_commission(PerOrder(buy_cost=0.0003, sell_cost=0.0013, min_cost=5))
    # 预测数据,通过options传入进来,使用 read_df 函数,加载到内存 (DataFrame)
    # 设置买入的股票数量,这里买入预测股票列表排名靠前的5只
    stock_count = 5
    # 每只的股票的权重,如下的权重分配会使得靠前的股票分配多一点的资金,[0.339160, 0.213986, 0.169580, ..]
    context.stock_weights = T.norm([1 / math.log(i + 2) for i in range(0, stock_count)])
    # 设置每只股票占用的最大资金比例
    context.max_cash_per_instrument = 0.2

# 回测引擎:每日数据处理函数,每天执行一次
def handle_data(context, data):
    # 按日期过滤得到今日的预测数据
    ranker_prediction = context.ranker_prediction[
        context.ranker_prediction.date == data.current_dt.strftime('%Y-%m-%d')]

    # 1. 资金分配
    # 平均持仓时间是hold_days,每日都将买入股票,每日预期使用 1/hold_days 的资金
    # 实际操作中,会存在一定的买入误差,所以在前hold_days天,等量使用资金;之后,尽量使用剩余资金(这里设置最多用等量的1.5倍)
    is_staging = context.trading_day_index < context.options['hold_days'] # 是否在建仓期间(前 hold_days 天)
    cash_avg = context.portfolio.portfolio_value / context.options['hold_days']
    cash_for_buy = min(context.portfolio.cash, (1 if is_staging else 1.5) * cash_avg)
    cash_for_sell = cash_avg - (context.portfolio.cash - cash_for_buy)
    positions = {e.symbol: p.amount * p.last_sale_price
                 for e, p in context.perf_tracker.position_tracker.positions.items()}

    # 2. 生成卖出订单:hold_days天之后才开始卖出;对持仓的股票,按StockRanker预测的排序末位淘汰
    if not is_staging and cash_for_sell > 0:
        equities = {e.symbol: e for e, p in context.perf_tracker.position_tracker.positions.items()}
        instruments = list(reversed(list(ranker_prediction.instrument[ranker_prediction.instrument.apply(
                lambda x: x in equities and not context.has_unfinished_sell_order(equities[x]))])))
        # print('rank order for sell %s' % instruments)
        for instrument in instruments:
            context.order_target(context.symbol(instrument), 0)
            cash_for_sell -= positions[instrument]
            if cash_for_sell <= 0:
                break

    # 3. 生成买入订单:按StockRanker预测的排序,买入前面的stock_count只股票
    buy_cash_weights = context.stock_weights
    buy_instruments = list(ranker_prediction.instrument[:len(buy_cash_weights)])
    max_cash_per_instrument = context.portfolio.portfolio_value * context.max_cash_per_instrument
    for i, instrument in enumerate(buy_instruments):
        cash = cash_for_buy * buy_cash_weights[i]
        if cash > max_cash_per_instrument - positions.get(instrument, 0):
            # 确保股票持仓量不会超过每次股票最大的占用资金量
            cash = max_cash_per_instrument - positions.get(instrument, 0)
        if cash > 0:
            context.order_value(context.symbol(instrument), cash)


# 调用交易引擎
m6 = M.trade.v2(
    instruments=None,
    start_date=conf.split_date,
    end_date=conf.end_date,
    prepare=prepare,
    initialize=initialize,
    handle_data=handle_data,
    order_price_field_buy='open',       # 表示 开盘 时买入
    order_price_field_sell='close',     # 表示 收盘 前卖出
    capital_base=1000000,               # 初始资金
    benchmark='000300.SHA',             # 比较基准,不影响回测结果
    # 通过 options 参数传递预测数据和参数给回测引擎
    options={'hold_days': conf.hold_days, 'model_id': m5.model_id}
)
[2017-09-04 09:04:07.096339] INFO: bigquant: advanced_auto_labeler.v1 开始运行..
[2017-09-04 09:04:07.099219] INFO: bigquant: 命中缓存
[2017-09-04 09:04:07.100050] INFO: bigquant: advanced_auto_labeler.v1 运行完成[0.003737s].
[2017-09-04 09:04:07.105973] INFO: bigquant: general_feature_extractor.v5 开始运行..
[2017-09-04 09:04:07.107814] INFO: bigquant: 命中缓存
[2017-09-04 09:04:07.108526] INFO: bigquant: general_feature_extractor.v5 运行完成[0.00257s].
[2017-09-04 09:04:07.113668] INFO: bigquant: derived_feature_extractor.v2 开始运行..
[2017-09-04 09:04:07.115656] INFO: bigquant: 命中缓存
[2017-09-04 09:04:07.116416] INFO: bigquant: derived_feature_extractor.v2 运行完成[0.002747s].
[2017-09-04 09:04:07.121714] INFO: bigquant: transform.v2 开始运行..
[2017-09-04 09:04:07.123620] INFO: bigquant: 命中缓存
[2017-09-04 09:04:07.124353] INFO: bigquant: transform.v2 运行完成[0.002635s].
[2017-09-04 09:04:07.129463] INFO: bigquant: join.v2 开始运行..
[2017-09-04 09:04:07.131535] INFO: bigquant: 命中缓存
[2017-09-04 09:04:07.132249] INFO: bigquant: join.v2 运行完成[0.002785s].
[2017-09-04 09:04:07.137976] INFO: bigquant: stock_ranker_train.v4 开始运行..
[2017-09-04 09:04:07.139721] INFO: bigquant: 命中缓存
[2017-09-04 09:04:07.140765] INFO: bigquant: stock_ranker_train.v4 运行完成[0.002774s].
[2017-09-04 09:04:07.159288] INFO: bigquant: backtest.v7 开始运行..
[2017-09-04 09:04:07.161076] INFO: bigquant: 命中缓存
/var/app/enabled/pandas/core/indexing.py:141: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
  self._setitem_with_indexer(indexer, value)
  • 收益率1.51%
  • 年化收益率0.77%
  • 基准收益率-6.33%
  • 阿尔法0.03
  • 贝塔0.89
  • 夏普比率-0.08
  • 收益波动率32.46%
  • 信息比率0.25
  • 最大回撤49.72%
[2017-09-04 09:04:08.314007] INFO: bigquant: backtest.v7 运行完成[1.154674s].
In [ ]:
 

(qwe9527) #3

原来表达式引擎构建因子要用训练函数M.stock_ranker_train的最新版本,用v3版本就会报错。。。。。


(小Q) #4

嗯。
不仅如此,你还可以对比以下几个模块:

M.advanced_auto_labeler.v1
M.general_feature_extractor.v5
M.derived_feature_extractor.v2
M.transform.v2

由于本例只是验证策略跑不跑的通,因此只拿了300只股票来举例。


(maida) #5

老哥厉害啊,哪里人?


(qwe9527) #6

江湖人。。


(maida) #7

都是江湖人啊,哈哈哈哈,有空喝喝茶