Matrix multiplication between pandas dataframe

user_16214945 0
Viewed 103

I am attempting to find dot product between a dataframe and a column vector (another dataframe). The dataframe looks like below:

df = pd.DataFrame([
    [0,23,0, 0, 1,0],
    [1,33,1, 0, 0,0],
    [2,40,1, 0, 1,1]],
    columns=['SN','Age',  'Nice_540', 'Nice_200', 'Nice_153','Nice_124'])

SN   |Age |Nice_540 |Nice_200| Nice_153|Nice_124|
----|-----|--------|--------|----------|--------|
0   |  23 |    0   |    0   |     1    |   0    |
1   |  33 |    1   |    0   |     0    |   0    |
2   |  40 |    1   |    0   |     1    |   1    |

I want the rows to matrix multiply numbers of last four column headings (540; 200, etc) so that the last column "FINAL" looks like this after matrix multiplication:

SN  | Age |Nice_540 |Nice_200| Nice_153|Nice_124|FINAL   |
----|-----|--------|--------|----------|--------|--------|
0   |  23 |    0   |    0   |     1    |   0    |   153  |
1   |  33 |    1   |    0   |     0    |   0    |   540  |
2   |  40 |    1   |    0   |     1    |   1    |   817  |

I have got the first two numbers in the last column right (153 and 540); but for the last one I am getting 540153124 and not 817 as would be expected from dot product.

This is what I have done so far:

AAAA = df.columns.to_list()

AAAA = str(AAAA)
string_pattern = r"\d{3}"
regex_pattern = re.compile(string_pattern)
BBBB = regex_pattern.findall(AAAA)

cols=df.filter(regex='Nice_',axis=1).columns

seR = pd.DataFrame(BBBB,index=cols)

df["FINAL"] = df[cols].dot(seR)

What am I doing wrong to get 540153124 and not 817? Thanks in advance.

dot-product pandas matrix-multiplication dataframe
1 Answers

You could do:

d = df.filter(like='Nice_')

vals = d.columns.str.split('_').str[-1].astype(int)

df['FINAL'] = d.dot(vals)

output:

   SN  Age  Nice_540  Nice_200  Nice_153  Nice_124  FINAL
0   0   23         0         0         1         0    153
1   1   33         1         0         0         0    540
2   2   40         1         0         1         1    817
user_163434640
Related