Question

我想要使用numpy和pandas作为我的依赖项来创建神经网络的问题。网络应该以日期，时间，纬度和经度为特征来预测地震的震级。这是数据集中的片段：

Date    Time    Latitude    Longitude   Magnitude
0   01/02/1965  13:44:18    19.246  145.616 6.0
1   01/04/1965  11:29:49    1.863   127.352 5.8
2   01/05/1965  18:05:58    -20.579 -173.972    6.2
3   01/08/1965  18:49:43    -59.076 -23.557 5.8
4   01/09/1965  13:32:50    11.938  126.427 5.8

这里的代码：

import pandas as pd
import numpy as np

data = pd.read_csv("C:/Users/Kamalov/AppData/Local/Programs/Python/Python35/my_code/datasets/database.csv")
train, test = data[:15000], data[15000:]
trainX, trainY = train[["Date","Time","Latitude","Longitude"]], train['Magnitude']
testX, testY = test[["Date","Time","Latitude","Longitude"]], test['Magnitude']

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

def sigmoid_output_to_derivative(output):
    return output*(1-output)

synapse_0 = 2*np.random.random((4,1)) - 1
synapse_1 = 2*np.random.random((1,4)) - 1

X = trainX.values
y = trainY.values


for iter in range(50000):
    # forward propagation
    layer_0 = X
    layer_1 = sigmoid(np.dot(layer_0,synapse_0))

    layer_2 = sigmoid(np.dot(layer_1,synapse_1))

    # how much did we miss?
    layer_2_error = layer_2 - y

    # multiply how much we missed by the 
    # slope of the sigmoid at the values in l1
    layer_2_delta = layer_2_error * sigmoid_output_to_derivative(layer_2)
    synapse_0_derivative = np.dot(layer_0.T,layer_2_delta)

    # update weights
    synapse_0 -= synapse_0_derivative

print ("Output After Training:")
print (layer_2)

我正在

＆＃34;不能将序列乘以非int类型＆＃39; float＆＃39;＆＃34;

错误，即使我将数据帧转换为numpy数组。

感谢任何帮助：/

Answer 1

错误消息可能有点误导。这是因为您的DataFrame包含dtype object列，在您的情况下包含日期和时间列。转换为numpy ndarray并不会有太大帮助，因为数据类型不会改变。您需要先将这些列转换为int或float值，然后才能使用np.dot()。

神经网络 - TypeError：不能将序列乘以非类型的＆int;＆＃39; float＆＃39;

1 个答案: