끝없는 인터넷 검색은 나를 파이썬에 대해 더 잘 교육 받고 질책을 느끼게했지만 여전히 내 작업을 해결하는 데는 단서가 없습니다. 정수/부동 소수점 값의 CSV를 읽고 신경망을 사용하여 값을 예측하고 싶습니다. 아이리스 데이터 집합을 읽고 분류를하는 몇 가지 예를 발견했지만 회귀 분석을 위해 어떻게 작동하는지 이해하지 못합니다. 누군가 내가 점들을 연결하도록 도울 수 있습니까? 925 개 값이어야CSV >> Tensorflow >> 회귀 (신경망을 통한) 모델
16804,0,1,0,1,1,0,1,0,1,0,1,0,0,1,1,0,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,1,0,0,1,1,0,0,1,0,1,0,1,0,1,0,1,0,1,0,1,1,0,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,1,0,0,1,0,1,0,1,0,1,1,0,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,1,0,0,1,0,1,0,1,0,1,0,1,0,1,1,0,0,1,0,0,0,1,1,0,0,1,0,0,0,0,0,1,0,1,0,0,0,0,0,1,0,1,0,0,0,0,0,1,0,0,0,1,0,0,0,0,1,0,0,1,0,0,0,1,0,0,0,1,0,0,0,1,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,1,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,1,0,0,1,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.490265,0.620805,0.54977,0.869299,0.422268,0.351223,0.33572,0.68308,0.40455,0.47779,0.307628,0.301921,0.318646,0.365993,6135.81
:
여기에 입력 한 선이다. 마지막 열은 출력입니다. 첫 번째는 RowID입니다. 이미 하나의 핫 인코딩을했기 때문에 대부분 이진 값입니다. 테스트 파일에는 출력/마지막 열이 없습니다. 전체 교육 파일에는 약 10 백만 개의 행이 있습니다. 일반적인 MxN 솔루션이 가능합니다.
편집 : 아이리스가 분류 문제이기 때문에이 샘플 데이터를 사용하겠습니다.하지만 위의 내용이 실제 목표입니다. ID 열을 제거했습니다. 6 개의 다른 열이 주어진 마지막 열을 예측해 봅시다. 여기에는 45 개의 행이 있습니다. (SRC : http://www.stat.ufl.edu/~winner/data/civwar2.dat)
100,1861,5,2,3,5,38 112,1863,11,7,4,59.82,15.18 113,1862,34,32,1,79.65,2.65 90,1862,5,2,3,68.89,5.56 93,1862,14,10,4,61.29,17.2 179,1862,22,19,3,62.01,8.89 99,1861,22,16,6,67.68,27.27 111,1862,16,11,4,78.38,8.11 107,1863,17,11,5,60.75,5.61 156,1862,32,30,2,60.9,12.82 152,1862,23,21,2,73.55,6.41 72,1863,7,3,3,54.17,20.83 134,1862,22,21,1,67.91,9.7 180,1862,23,16,4,69.44,3.89 143,1863,23,19,4,81.12,8.39 110,1862,16,12,2,31.82,9.09 157,1862,15,10,5,52.23,24.84 101,1863,4,1,3,58.42,18.81 115,1862,14,11,3,86.96,5.22 103,1862,7,6,1,70.87,0 90,1862,11,11,0,70,4.44 105,1862,20,17,3,80,4.76 104,1862,11,9,1,29.81,9.62 102,1862,17,10,7,49.02,6.86 112,1862,19,14,5,26.79,14.29 87,1862,6,3,3,8.05,72.41 92,1862,4,3,0,11.96,86.96 108,1862,12,7,3,16.67,25 86,1864,0,0,0,2.33,11.63 82,1864,4,3,1,81.71,8.54 76,1864,1,0,1,48.68,6.58 79,1864,0,0,0,15.19,21.52 85,1864,1,1,0,89.41,3.53 85,1864,1,1,0,56.47,0 85,1864,0,0,0,31.76,15.29 87,1864,6,5,0,81.61,3.45 85,1864,5,5,0,72.94,0 83,1864,0,0,0,46.99,2.38 101,1864,5,5,0,1.98,95.05 99,1864,6,6,0,42.42,9.09 10,1864,0,0,0,50,9 98,1864,6,6,0,79.59,3.06 10,1864,0,0,0,71,9 78,1864,5,5,0,70.51,1.28 89,1864,4,4,0,59.55,13.48
날이 일반적인 작업입니다,하지만이를 요청했습니다 내가 이렇게 읽은 어떤 포럼에서 답변을 얻지 못한 것으로 보인다 것을 추가 할 수 있습니다. 내가 깨뜨린 코드를 줄 수도 있지만, 기능적으로 올바르지 않은 코드로 시간을 낭비하고 싶지는 않습니다. 미안해, 내가 이런 식으로 물어 봤어. API를 이해하지 못하고 설명서에서 데이터 유형을 알지 못합니다.
#!/usr/bin/env python
import tensorflow as tf
import csv
import numpy as np
from numpy import genfromtxt
# Build Example Data is CSV format, but use Iris data
from sklearn import datasets
from sklearn.cross_validation import train_test_split
import sklearn
def buildDataFromIris():
iris = datasets.load_iris()
data = np.loadtxt(open("t100.csv.out","rb"),delimiter=",",skiprows=0)
labels = np.copy(data)
labels = labels[:,924]
print "labels: ", type (labels), labels.shape, labels.ndim
data = np.delete(data, [924], axis=1)
print "data: ", type (data), data.shape, data.ndim
그리고 여기 내가 사용하고자하는 기본 코드입니다 : 여기
내가 그 두 ndarrays으로 CSV를 읽고있는 최신 코드입니다. 이 예가 나온 것도 아니 었습니다. 아래의 링크에있는 API는 모호합니다. 적어도 DNNRegressor에 입력 된 데이터 유형과 문서의 다른 데이터 유형을 파악할 수 있다면 사용자 정의 코드를 작성할 수 있습니다.
estimator = DNNRegressor(
feature_columns=[education_emb, occupation_emb],
hidden_units=[1024, 512, 256])
# Or estimator using the ProximalAdagradOptimizer optimizer with
# regularization.
estimator = DNNRegressor(
feature_columns=[education_emb, occupation_emb],
hidden_units=[1024, 512, 256],
optimizer=tf.train.ProximalAdagradOptimizer(
learning_rate=0.1,
l1_regularization_strength=0.001
))
# Input builders
def input_fn_train: # returns x, Y
pass
estimator.fit(input_fn=input_fn_train)
def input_fn_eval: # returns x, Y
pass
estimator.evaluate(input_fn=input_fn_eval)
estimator.predict(x=x)
그리고 나서 큰 문제는 이들이 함께 작동하는 것입니다.
다음은 내가 본 몇 페이지입니다. https://www.tensorflow.org/versions/r0.11/tutorials/tflearn/index.html
- 자료 코드 https://www.tensorflow.org/versions/r0.11/api_docs/python/contrib.learn.html#DNNRegressor -CSV 읽기 : https://www.tensorflow.org/versions/master/how_tos/reading_data/index.html#csv-files
- 열 삽입 : API를 https://www.tensorflow.org/versions/r0.11/tutorials/wide_and_deep/index.html
- 목록 (DNNRegressor, TensorFlowDNNRegressor, LinearRegressor, TensorFlowLinearRegressor, TensorFlowRNNRegressor, TensorFlowRe gressor) : https://www.tensorflow.org/versions/r0.11/api_docs/python/contrib.learn.html
문제 설명을 좁힐 수 있습니까? 현재의 문제는 (1) 한 번 핫 형식으로 CSV 파일을 섭취하는 방법; (2) 엄격한 분류 (열거 유형)에서 채점 (CSV 라인 끝의 부동 값)으로 전환하는 방법. – Prune
변환 시도와 결과 (출력 부족)를 게시 할 수 있습니까? 그것은 우리에게 공격에 더욱 집중된 문제를 줄 것입니다. – Prune
본질적으로 CSV를 읽을 수 있기 때문에 범위를 단순화했으나 원하는 형식으로 NN에 입력하는 방법을 모르겠습니다. – ldmtwo