我需要用Panda阅读这个csv file,对它执行一些处理,然后将剩余的10%的数据写入另一张表。
鉴于此解决方案(https://stackoverflow.com/a/55763598/3373710),我想在取出10%行后对store_data的其余部分执行一个处理,但是,elif条件打印原始文件的相同行,如何修复我的条件以跳过10%行?
store_data = pd.read_csv("heart_disease.csv")
with open("out1.csv","w") as outfile:
outcsv = csv.writer(outfile)
for i, row in store_data.iterrows():
if not i % 10: #write 10% of the file to another file
outcsv.writerow(row)
elif i % 10: #I need to do some process on the rest of the file
store_data = store_data.applymap(str)
简单地将数据框分成两部分,将10%保存到文件(dataframe.to_csv(..)
)并将计算应用到第二个df中的90%更简单,更简洁。
您可以通过计算一个新列来告诉您行是否为test,并将此数据帧分成两个新列值:
数据文件创建:
fn = "heart_disease.csv"
with open(fn,"w") as f:
# doubled the data provided
f.write("""Age,AL,SEX,DIAB,SMOK,CHOL,LAD,RCA,LM
65,0,M,n,y,220,80,75,20\n45,0.2,F,n,n,300,90,35,35\n66,-1,F,y,y,200,90,80,20
70,0.2,F,n,y,220,40,85,15\n80,1.1,M,y,y,200,90,90,25\n55,0,M,y,y,240,95,45,25
90,-1,M,n,y,350,35,75,20\n88,1,F,y,y,200,40,85,20\n50,1.1,M,n,n,220,55,30,30
95,-1,M,n,y,230,75,85,15\n30,1.1,F,n,y,235,75,20,30
65,0,M,n,y,220,80,75,20\n45,0.2,F,n,n,300,90,35,35\n66,-1,F,y,y,200,90,80,20
70,0.2,F,n,y,220,40,85,15\n80,1.1,M,y,y,200,90,90,25\n55,0,M,y,y,240,95,45,25
90,-1,M,n,y,350,35,75,20\n88,1,F,y,y,200,40,85,20\n50,1.1,M,n,n,220,55,30,30
95,-1,M,n,y,230,75,85,15\n30,1.1,F,n,y,235,75,20,30
""")
程序:
import pandas as pd
fn = "heart_disease.csv"
store_data = pd.read_csv(fn)
print(store_data)
import random
import numpy as np
percentage = 0.1
store_data["test"] = np.random.rand(len(store_data))
test_data = store_data[store_data.test <= percentage]
other_data = store_data[store_data.test > percentage]
print(test_data)
print(other_data)
输出:
# original data
Age AL SEX DIAB SMOK CHOL LAD RCA LM
0 65 0.0 M n y 220 80 75 20
1 45 0.2 F n n 300 90 35 35
2 66 -1.0 F y y 200 90 80 20
3 70 0.2 F n y 220 40 85 15
4 80 1.1 M y y 200 90 90 25
5 55 0.0 M y y 240 95 45 25
6 90 -1.0 M n y 350 35 75 20
7 88 1.0 F y y 200 40 85 20
8 50 1.1 M n n 220 55 30 30
9 95 -1.0 M n y 230 75 85 15
10 30 1.1 F n y 235 75 20 30
11 65 0.0 M n y 220 80 75 20
12 45 0.2 F n n 300 90 35 35
13 66 -1.0 F y y 200 90 80 20
14 70 0.2 F n y 220 40 85 15
15 80 1.1 M y y 200 90 90 25
16 55 0.0 M y y 240 95 45 25
17 90 -1.0 M n y 350 35 75 20
18 88 1.0 F y y 200 40 85 20
19 50 1.1 M n n 220 55 30 30
20 95 -1.0 M n y 230 75 85 15
21 30 1.1 F n y 235 75 20 30
# data with test <= 0.1
Age AL SEX DIAB SMOK CHOL LAD RCA LM test
3 70 0.2 F n y 220 40 85 15 0.093135
10 30 1.1 F n y 235 75 20 30 0.021302
# data with test > 0.1
Age AL SEX DIAB SMOK CHOL LAD RCA LM test
0 65 0.0 M n y 220 80 75 20 0.449546
1 45 0.2 F n n 300 90 35 35 0.953321
2 66 -1.0 F y y 200 90 80 20 0.928233
4 80 1.1 M y y 200 90 90 25 0.672880
5 55 0.0 M y y 240 95 45 25 0.136537
6 90 -1.0 M n y 350 35 75 20 0.439261
7 88 1.0 F y y 200 40 85 20 0.935340
8 50 1.1 M n n 220 55 30 30 0.737416
9 95 -1.0 M n y 230 75 85 15 0.461699
11 65 0.0 M n y 220 80 75 20 0.548624
12 45 0.2 F n n 300 90 35 35 0.679861
13 66 -1.0 F y y 200 90 80 20 0.195141
14 70 0.2 F n y 220 40 85 15 0.997854
15 80 1.1 M y y 200 90 90 25 0.871436
16 55 0.0 M y y 240 95 45 25 0.907141
17 90 -1.0 M n y 350 35 75 20 0.295690
18 88 1.0 F y y 200 40 85 20 0.970249
19 50 1.1 M n n 220 55 30 30 0.566218
20 95 -1.0 M n y 230 75 85 15 0.545188
21 30 1.1 F n y 235 75 20 30 0.217490
它是随机的,您可能会获得10%的数据 - 或者您可以获得更少/超过10% - 数据越大,您就越接近10%。
您可以使用“派生”数据帧使用df.to_csv
将数据存储到测试和其他数据中。
对于一个纯粹的熊猫解决方案How do I create test and train samples from one dataframe with pandas?是你的副本,但你似乎单独处理csv所以不确定它是否适用。
这是一个纯粹的熊猫解决方案:
import pandas as pd
df = pd.read_csv("heart_disease.csv")
#select only 10% of the rows, subtract 1 because index starts with zero
df_slice = df.loc[:round(len(df) * 10 /100) - 1, :]
#write the sliced df to csv
df_slice.to_csv("sliced.csv", index=None)
#to work with the rest of the data, just drop the rows at index where the df_slice rows exist
l = df_slice.index.tolist()
df.drop(df.index[l], inplace=True) #90% of data
#now the df has the rest 90% and you can do whatever you want with it