如何在 Python 中导入这个数据文件？

Question

So I just started with Python and I have no idea how to import the text file the way I want it to.

This is my code so far:

f = open("Data.txt", "r")
attributes = f.readlines()

w1 = 2 * np.random.random((1,19)) -1

def sigmoid(x):
    return 1 / (1 + np.exp(-x))

outputs = sigmoid(np.dot(attributes, w1))

So the problem here is that I get the error message:

Cannot cast array data from dtype('float64') to dtype('<U32') according to the rule 'safe'.

I know that the problem is that the code does not read the text file as an array of numbers which is why I get the error.

This is one line in the text file:

1,1,22,22,22,19,18,14,49.895756,17.775994,5.27092,0.771761,0.018632,0.006864,0.003923,0.003923,0.486903,0.100025,1,0

Answer 1

You're almost there :

import numpy as np
with open("Data.txt", "r") as f:
    attributes = f.readlines()
attributes = [row.strip().split(",") for row in attributes]
attributes = [[float(x) for x in row] for row in attributes]

w1 = 2 * np.random.random((1,20)) -1

def sigmoid(x):
    return 1 / (1 + np.exp(-x))

outputs = sigmoid(np.dot(np.array(attributes), w1.T))

You needed something like "list comprehension" to split values on each row (if there are more than one). Also, you were missing one element in w1 to match the dimension of attributes. And finally, you will have to use the transposition of w1 to make the production using np.dot (np.array(attributes)'s shape is (1,20) as well as w1's.

Also : always remember to use this "with" statement to open files (it automatically close the file after ending the statement ; otherwise, you might get into some trouble...)

---

Edit

If your real dataset contains more data, you should consider the use of pandas, which will be much more efficient :

import pandas as pd
df = pd.read_csv("Data.txt"), sep=",", header=None)
numpy_array = df.values

Answer 2

I think what you want to do is something like the following:

import numpy as np


f = open("Data.txt", "r")

# Read the file, replace newline by "", split the string on commas
attributes = f.read().replace("\n", "").split(",")

# convert every element from string to float
attributes = np.array([float(element) for element in attributes])

# 20 random values in [-1,1)
w1 = 2 * np.random.random(20) - 1

def sigmoid(x):
    return 1 / (1 + np.exp(-x))

outputs = sigmoid(np.dot(attributes, w1))

# Printing results
print(attributes)
print(w1)
print(outputs)