将R语言dplyr代码转换为Python dfply代码报错求助

Let's break down what's going wrong with your code and fix it step by step. Your error AttributeError: 'DataFrame' object has no attribute 'invoiceprob' happens because you're trying to access the invoiceprob column from the original test DataFrame, but this column is only created later in the summarize step of your pipeline. Plus, there are a couple of other syntax issues that don't align with your original R dplyr logic.

What's Wrong with Your Original Code? #

1. Referencing the original DataFrame instead of pipeline data: When you use test.InvoiceDocNumber or test.itemprob, you're pulling directly from the original test DataFrame, not the data being passed through the pipeline. This causes two big problems:
   • max(test.itemprob) calculates the maximum value of the entire itemprob column across all rows, not the maximum per group (which is what your R code does).
   • test.invoiceprob doesn't exist—this column is created in the summarize step, so the original DataFrame has no idea about it.
2. Incorrect ranking approach: Using rankdata(test.invoiceprob) again references the original DataFrame (which lacks the column) and ignores the pipeline's current state.

Corrected dfply Code #

Here's the fixed version that matches your original R dplyr logic exactly:

from dfply import *

# Option 1: Using pandas' built-in rank method (most similar to R's rank)
tmp = (test 
       >> group_by(X.InvoiceDocNumber) 
       >> summarize(invoiceprob=X.itemprob.max()) 
       >> mutate(invoicerank=X.invoiceprob.rank(ascending=False)))

Or if you prefer using rankdata from scipy (note the negative sign to replicate descending ranking):

from dfply import *
from scipy.stats import rankdata

tmp = (test 
       >> group_by(X.InvoiceDocNumber) 
       >> summarize(invoiceprob=X.itemprob.max()) 
       >> mutate(invoicerank=rankdata(-X.invoiceprob)))

Key Explanations #

• The X symbol: This is dfply's way of referring to the current DataFrame being passed through the pipeline. It's equivalent to the implicit data reference in R's dplyr. Using X.InvoiceDocNumber, X.itemprob, and X.invoiceprob ensures you're always working with the data at each step of the pipeline.
• Group-wise max: X.itemprob.max() calculates the maximum itemprob for each group defined by InvoiceDocNumber, just like max(itemprob) in your R code.
• Descending ranking:
  • X.invoiceprob.rank(ascending=False) replicates R's rank(desc(invoiceprob))—it ranks values from highest to lowest.
  • If using rankdata, adding a negative sign (-X.invoiceprob) converts descending ranking into ascending (since rankdata defaults to ascending order).

Quick Comparison to Your Original R Code #

This fix will resolve the AttributeError and ensure your Python code behaves exactly like your original R code.

内容的提问来源于stack exchange，提问作者vinay karagod