How to create a new dataframe with unique value on lesser rows and summarize result?

So i have this dataframe df that have 3 column of unique data and the rest are result column with no headers. The ID will always tie to the item and cost, for example, apple will always have the ID of 12 and cost of 5.

print(df)
    ID   Item   cost   3   4  
0   12   Apple   5     Y   N
1   12   Apple   5     N   N
2   12   Apple   5     Y   N
3   12   Apple   5     Y   Y
4   15   Orange  6     N   Y
5   15   Orange  6     N   N
6   15   Orange  6     N   Y
7   15   Orange  6     N   Y
8   21   Lemon   6     Y   NaN
9   51   Grape   6     Y   N
10  21   Lemon   6     Y   NaN

I want to create a similar dataframe but this dataframe must have the unique value of Item as the maximum number of row. And the rest of column must count the number of Y and N to to put into the column. If 1 Y appears in column 3 on apple, the result will be Y. Only all of value is N, then apple will be N. If there is only NaN, it will return NaN.

Here is my expected dataframe to look like, df2

print(df2)
    ID   Item   cost   3   4  
0   12   Apple   5     Y   Y
1   15   Orange  6     N   Y
2   21   Lemon   6     Y   NaN
3   51   Grape   6     Y   N

I use this code to try to create df2. But it give KeyError: False on the YCount function.

df2 = df.drop_duplicates(subset=['ID', 'Item'], keep='first')   
#copy df to df2 with ID, Item, cost duplicates removed
df2[df2.columns[3:]] = ''
#clear column 3 to 4

for i in df2["Item"].unique():
    for x in range(3, len(df2.columns)):
        YCount =(df["Item" == i].df.iloc[:,x] == 'Y').sum()    #count number of Y corresponding to the item
        NCount =(df["Item" == i].df.iloc[:,x] == 'N').sum()    #count number of N corresponding to the item
        if YCount > 0:
            df2.iloc[:,x] = "Y"                                #if more than zero Y appears, put Y
        elif YCount + NCount == 0:
            df2.iloc[:,x] = ""                                 #if total Y and N is 0, put NaN
        elif YCount == 0 and NCount > 0:
            df2.iloc[:,x] = "N"                                #if Y=0 and N more than 0, put N

What you want to do is very easy with booleans, so just replace 'Y' by True and 'N' by False, the perform a groupby.max:

(df
 .replace({'Y': True, 'N': False})
 .groupby(['ID', 'Item', 'cost'], as_index=False)
 .max()
 .replace({True: 'Y', False: 'N'})
)

Note that 'Y' is lexicographically sorted after 'N', so you could also use:

out = df.groupby(['ID', 'Item', 'cost'], as_index=False).max()

but this only works with these particular values. This will also be less robust if you have a mix of 'Y'/'N'/NaN.

Output:

   ID    Item  cost  3    4
0  12   Apple     5  Y    Y
1  15  Orange     6  N    Y
2  21   Lemon     6  Y  NaN
3  51   Grape     6  Y    N