预分配对象大小

Question

我有这个脚本来计算交易的盈利和亏损。它工作正常但我认为它可以改进。摆脱for循环至少可以使代码看起来很紧凑。有人可以帮帮我吗？

计算盈利/亏损的逻辑首先是将卖出交易与潜在买入交易相匹配。单个卖出交易可以与多个买入匹配。因此，成本可能会分配给多个购买。

步骤：

将交易分为买入和卖出日期。
计算平均成本价格
计算利润/亏损=（销售价格 - 成本价格）*匹配vol

由于

以下是样本数据集

> structure(list(AsxCode = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), .Label = "QAN", class = "factor"), Order.Type = structure(c(1L, 2L, 2L, 1L, 1L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 2L, 1L), .Label = c("Buy", "Sell"), class = "factor"), Trade.Date = structure(c(13L, 12L, 12L, 11L, 10L, 9L, 8L, 7L, 6L, 5L, 4L, 3L, 2L, 1L), .Label = c("2014-03-28", "2014-05-22", "2014-11-07", "2014-11-18", "2014-12-04", "2015-03-02", "2015-03-24", "2015-03-27", "2015-05-11", "2015-05-15", "2015-08-21", "2016-04-15", "2016-04-18"), class = "factor"), Price = c(3.75, 4.05, 4.01, 3.55, 3.68, 3.38, 2.9, 2.98, 2.9, 2.05, 1.8, 1.65, 1.25, 1.07), Quantity = c(850L, 1350L, 150L, 1000L, 1500L, 1400L, 1091L, 2000L, 1750L, 600L, 366L, 375L, 500L, 500L), Consideration = c(3198.5, 5456.5, 590.5, 3561, 5531, 4743, 3152.9, 5949, 5086, 1241, 669.8, 629.75, 614, 546), match_status = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA), match_vol = c(0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L), avg_price = c(0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L), profit_loss = c(0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L)), .Names = c("AsxCode", "Order.Type", "Trade.Date", "Price", "Quantity", "Consideration", "match_status", "match_vol", "avg_price", "profit_loss"), row.names = c(NA, -14L), class = "data.frame")

   AsxCode Order.Type Trade.Date Price Quantity Consideration match_status match_vol avg_price profit_loss
1      QAN        Buy 2016-04-18  3.75      850       3198.50           NA         0         0           0
2      QAN       Sell 2016-04-15  4.05     1350       5456.50           NA         0         0           0
3      QAN       Sell 2016-04-15  4.01      150        590.50           NA         0         0           0
4      QAN        Buy 2015-08-21  3.55     1000       3561.00           NA         0         0           0
5      QAN        Buy 2015-05-15  3.68     1500       5531.00           NA         0         0           0
6      QAN        Buy 2015-05-11  3.38     1400       4743.00           NA         0         0           0
7      QAN       Sell 2015-03-27  2.90     1091       3152.90           NA         0         0           0
8      QAN       Sell 2015-03-24  2.98     2000       5949.00           NA         0         0           0
9      QAN        Buy 2015-03-02  2.90     1750       5086.00           NA         0         0           0
10     QAN        Buy 2014-12-04  2.05      600       1241.00           NA         0         0           0
11     QAN        Buy 2014-11-18  1.80      366        669.80           NA         0         0           0
12     QAN        Buy 2014-11-07  1.65      375        629.75           NA         0         0           0
13     QAN       Sell 2014-05-22  1.25      500        614.00           NA         0         0           0
14     QAN        Buy 2014-03-28  1.07      500        546.00           NA         0         0           0


calculate.profit <- function(trades){       
    trades$match_vol <- 0
    s <- trades[trades$Order.Type== 'Sell', ]
    sell.trades <- s[order(s$Trade.Date, decreasing=FALSE),]    

    b <- trades[trades$Order.Type== 'Buy', ]
    buy.trades <- b[order(b$Trade.Date, decreasing=FALSE),]     

    # Don't want to execute the for loop when there is no sell trades. In other words when there is no profit/loss unless you sell
    if(nrow(sell.trades)==0){
        return (buy.trades)
    }

    # for each sell find the associated buys
    for(i in 1:nrow(sell.trades))
    {           
        # calculate average price. The Consideration column contains total cost  
        s.price <- sell.trades[i, 'Consideration']/sell.trades[i,'Quantity']        

        for(j in 1:nrow(buy.trades))
        {   
            # this part matches sell with a buy trade
            # if sell volume and buy volume are same, the sell is fully matched otherwise it has to find the remaining sell units.      
            s.vol <- sell.trades[i,'Quantity'] - sell.trades[i,'match_vol']         
            b.vol <- buy.trades[j, 'Quantity'] - buy.trades[j, 'match_vol']

            if (b.vol != 0)         
            {               
                b.price <- buy.trades[j, 'Consideration']/buy.trades[j, 'Quantity']
                # contains the volume which is matched between buy and sell
                # trades
                match.vol <- min(s.vol, b.vol)              
                profit <- match.vol * (s.price - b.price)               

                buy.trades[j, 'match_vol'] <- match.vol + buy.trades[j, 'match_vol']

                sell.trades[i, 'profit_loss'] <- profit + sell.trades[i, 'profit_loss'] 
                sell.trades[i, 'match_vol'] <- match.vol + sell.trades[i, 'match_vol']              
            }

            # sell parcel fully processed           
            if (sell.trades[i ,'match_vol'] == sell.trades[i ,'Quantity'])
            {               
                j=1
                break;                   
            }                       
        }           
    }   
    return (rbind(buy.trades, sell.trades))
}

Answer 1

可以做出许多改进。

预分配对象大小

最明显的事情是预先分配对象尺寸。收到的智慧是inefficient to expand objects in loops。因此你会这样做：

# On example of a single column 
sell.trades.vec <- vector(mode = "numeric", length = nrow(buy.trades))

以避免对象在循环中消耗。

`seq_along()`

从广义上讲，使用seq_along代替1:something很简洁，看看：

>> a <- NULL
>> 1:length(a)
[1] 1 0
>> seq_along(a)
integer(0)
>>

：

>> 1:0
[1] 1 0
>> seq_along(0)
[1] 1
>>

我猜你会（很可能）总是有一些明智的nrow值，但seq_along可能值得反思，以防有可能得到一些奇怪的数据。< / p>

计算交易的盈亏

1 个答案:

预分配对象大小

`seq_along()`