擅长:python、mysql、java
<pre><code>import pandas as pd
a = pd.DataFrame([["user1","user10","user1"],
["user4","user10","user5"],
["user3","user3","user12"]] ,
columns=["i1","i2","i3"]) #simplified input dataframe
for i in a.index:
m = a.loc[i].duplicated() #mask array for each rows
a.loc[i] = a.loc[i].mask(m).fillna("") #filter duplicates and fill by empty string
</code></pre>
<p>我认为这个解决方案可以从性能的角度进行改进,但它是有效的。你知道吗</p>