CODEWITHSUNDEEP
pythonmatrix
Viki
Viki

Reputation: 101

Create a matrix from a text file - python

I would like to create a matrix from a three column file. I am sure it's something extremely easy, but I just do not understand how it needs to be done. Please be gentle, I am a beginner to python. Thank you

The format of my input file

A A 5 
A B 4 
A C 3 
B B 2 
B C 1 
C C 0

Desired output - complete matrix

  A B C
A 5 4 3
B 4 2 1
C 3 1 0

Or - half matrix

  A B C
A 5 4 3
B   2 1
C     0

I tried this, but as I said, I am VERY new to python and programming.

import numpy as np

for line in file('test').readlines():
    name1, name2, value = line.strip().split('\t')

a = np.matrix([[name1], [name2], [value]])
print a

WORKING SCRIPT - One of my friend also helped me, so if anyone if interested in a simpler script, here it is. It's not the most efficient, but works perfectly.

data = {}
names = set([])

for line in file('test').readlines():
    name1, name2, value = line.strip().split('\t')
    data[(name1, name2)] = value
    names.update([name1])

names = sorted(list(names))
print  names
print data

output = open('out.txt', 'w')

output.write("\t%s\n" % ("\t".join(names)))
for nameA in names:
    output.write("%s" % nameA)
    for nameB in names:
        key = (nameA, nameB)
        if key in data:
            output.write("\t%s" % data[(nameA, nameB)]) 
        else:
            output.write("\t")  
    output.write("\n")


output.close()

Upvotes: 4

Views: 21803

Answers (4)

Moritz
Moritz

Reputation: 5418

Try:

import pandas as pd
import numpy as np

raw = []
with open('test.txt','r') as f:
    for line in f:
        raw.append(line.split())
data = pd.DataFrame(raw,columns = ['row','column','value'])
data_ind = data.set_index(['row','column']).unstack('column')
np.array(data_ind.values,dtype=float))

Output:

array([[ 5., 4., 3.], [ nan, 2., 1.], [ nan, nan, 0.]])

Upvotes: 4

ypx
ypx

Reputation: 1459

You're matrix seems to resember an adjacency matrix of a graph.

I find the answer with pandas much more concise and elegant. Here's my attempt without adding pandas as an additional dependency.

<!-- language: python -->
f = open('.txt', 'r');

EdgeKey = namedtuple("EdgeKey", ["src", "dst"])

g = dict()
for line in f:

    elems = line.split(' ');
    key = EdgeKey(src=elems[0], dst=elems[1])
    g[key] = elems[2]
    key_rev = EdgeKey(src=elems[1], dst=elems[0]) # g[A, B] == g[B, A]
    g[key_rev] = elems[2]

vertices = set()
for src, dst in g.keys():
    vertices.add(src)
    vertices.add(dst)

vertices = list(vertices)
vertices.sort()

# create adjacency matrix
mat  = np.zeros((len(vertices), len(vertices)))
for s, src in enumerate(vertices):
    for d, dst in enumerate(vertices):
        e = EdgeKey(src=src, dst=dst)
        if e in g:
            mat[s, d] = int(g[e])

# print adjacency matrix
print ' ' , ' '.join(vertices) # print header
for i, row in enumerate(mat):
    print vertices[i], ' '.join([str(int(c)) for c in row.tolist()])

Upvotes: 1

ljetibo
ljetibo

Reputation: 3094

Although there's already an accepted answer, it uses pandas. A relatively generic way of getting the same effect but by not using a additional library is this: (numpy is used because OP specified numpy, however you can achieve the same thing with lists)

import string
import numpy as np

up = string.ascii_uppercase
uppercase = list()
for letter in up:
    uppercase.append(letter)

file = open("a.txt")

matrix = np.zeros((3, 3))

for line in file.readlines():
    tmp = line.strip()
    tmp = tmp.split(" ")
    idx = uppercase.index(tmp[0])
    idy = uppercase.index(tmp[1])
    matrix[idx, idy] = tmp[2]

Idea is that you gather all the alphabetical letters, hopefully OP will limit themselves to just the English alphabet without special chars (šđćžčę°e etc...).

We create a list of from the alphabet so that we can use the index method to retrieve the index value. I.e. uppercase.index("A") is 0. We can use these indices to fill in our array.

Read in file line by line, strip extra characters, split by space to get:

['A', 'A', '5']
['A', 'B', '4']

This is now the actual working part:

    idx = uppercase.index(tmp[0])
    idy = uppercase.index(tmp[1])
    matrix[idx, idy] = tmp[2]

I.e. for letter "A", idx evaluates to 0 and so does idy. Then matrix[0,0] becomes the value tmp[2] which is 4. Following the same logic for "B" we get matrix[0,1]=5. And so on.

A more generalized case would be to declare matrix = np.zeros((3, 3)) as matrix = np.zeros((26, 26)) because there are 26 letters in english alphabet and the OP doesn't have to just use "ABC", but could potentially use the entire range A-Z.

Example output for upper program would be:

>>> matrix
array([[ 5.,  4.,  3.],
       [ 0.,  2.,  1.],
       [ 0.,  0.,  0.]])

Upvotes: 3

Rohit Barnwal
Rohit Barnwal

Reputation: 492

You can use this library http://docs.scipy.org/doc/numpy/reference/generated/numpy.matrix.html

You just need to make proper adjustment.

hope it helps.

Upvotes: 2

Related Questions