I'm trying to create a dataframe with timeseries data with multiple column headers, but I'm new to python and pandas. The data frame is one day of hourly data down and 1 million service points across. I can create the basic dataframe with the following:
intdata = pd.DataFrame(np.random.randint(0,1000,size=(24, 1000000)),
index=pd.date_range('2018-01-01',periods=24, freq='H'))
Index 0 1 2 3 4 5 6 7 ... 1000000
2018-01-01 00:00:00 458 352 905 176 801 438 193 172 123
2018-01-01 01:00:00 68 313 465 460 960 487 574 335 123
2018-01-01 02:00:00 769 984 253 143 592 741 709 660 123
2018-01-01 03:00:00 316 684 195 660 602 200 228 748 123
2018-01-01 04:00:00 201 947 514 696 12 288 577 957 123
2018-01-01 05:00:00 235 118 746 880 909 365 233 57 123
...
I'd like the index row to be followed by a set of characteristics for each point like:
Service Point 0 1 2 3 4 5 6 7 **** my index
Characteristic_1 A A C B A D B C **** characteristics needed
Characteristic_2 X Y Y Z Z J Q J
2018-01-01 00:00:00 458 352 905 176 801 438 193 172
2018-01-01 01:00:00 68 313 465 460 960 487 574 335
...
How do I do that?
Thanks
pandas.MultiIndex.from_tuples
You need to specify what the levels will be.
tups = [(0, 'A', 'X'), (1, 'A', 'Y'), (2, 'C', 'Y'), (3, 'B', 'Z')]
mcol = pd.MultiIndex.from_tuples(
tups, names=['Service Point', 'Characteristic_1', 'Characteristic_2'])
Then include these with the specification of the DataFrame
intdata = pd.DataFrame(
np.random.randint(0,1000,size=(10, 4)),
index=pd.date_range('2018-01-01',periods=10, freq='H'),
columns=mcol
)
intdata
Service Point 0 1 2 3
Characteristic_1 A A C B
Characteristic_2 X Y Y Z
2018-01-01 00:00:00 400 800 426 433
2018-01-01 01:00:00 920 123 250 113
2018-01-01 02:00:00 319 300 187 33
2018-01-01 03:00:00 673 230 696 472
2018-01-01 04:00:00 703 766 962 796
2018-01-01 05:00:00 322 295 414 734
2018-01-01 06:00:00 987 38 400 848
2018-01-01 07:00:00 350 275 494 833
2018-01-01 08:00:00 677 58 335 293
2018-01-01 09:00:00 284 195 742 355
If you have the levels in existing lists, you can use zip
chr1 = [*'AACBADBC']
chr2 = [*'XYYZZJQJ']
tups = [*zip(range(8), chr1, chr2)]
mcol = pd.MultiIndex.from_tuples(
tups, names=['Service Point', 'Characteristic_1', 'Characteristic_2'])
tidx = pd.date_range('2018-01-01',periods=10, freq='H')
data = np.random.randint(0, 1000, size=(len(tidx), len(mcol)))
intdata = pd.DataFrame(data, tidx, mcol)
intdata
Service Point 0 1 2 3 4 5 6 7
Characteristic_1 A A C B A D B C
Characteristic_2 X Y Y Z Z J Q J
2018-01-01 00:00:00 311 306 868 48 894 584 989 548
2018-01-01 01:00:00 848 170 592 640 638 400 112 642
2018-01-01 02:00:00 906 660 883 149 907 848 247 875
2018-01-01 03:00:00 461 432 479 733 979 540 311 86
2018-01-01 04:00:00 849 471 480 836 834 235 901 22
2018-01-01 05:00:00 758 193 45 405 739 818 81 577
2018-01-01 06:00:00 752 647 799 688 588 496 37 504
2018-01-01 07:00:00 380 785 750 975 960 535 971 257
2018-01-01 08:00:00 187 422 915 863 290 483 423 473
2018-01-01 09:00:00 270 144 749 710 983 755 839 709
pandas.MultiIndex.from_arrays
But then again, if you have the levels already in separate lists, you don't need to zip
them yourself
chr1 = [*'AACBADBC']
chr2 = [*'XYYZZJQJ']
mcol = pd.MultiIndex.from_arrays(
[range(8), chr1, chr2],
names=['Service Point', 'Characteristic_1', 'Characteristic_2'])
tidx = pd.date_range('2018-01-01',periods=10, freq='H')
data = np.random.randint(0, 1000, size=(len(tidx), len(mcol)))
intdata = pd.DataFrame(data, tidx, mcol)
intdata
Service Point 0 1 2 3 4 5 6 7
Characteristic_1 A A C B A D B C
Characteristic_2 X Y Y Z Z J Q J
2018-01-01 00:00:00 311 306 868 48 894 584 989 548
2018-01-01 01:00:00 848 170 592 640 638 400 112 642
2018-01-01 02:00:00 906 660 883 149 907 848 247 875
2018-01-01 03:00:00 461 432 479 733 979 540 311 86
2018-01-01 04:00:00 849 471 480 836 834 235 901 22
2018-01-01 05:00:00 758 193 45 405 739 818 81 577
2018-01-01 06:00:00 752 647 799 688 588 496 37 504
2018-01-01 07:00:00 380 785 750 975 960 535 971 257
2018-01-01 08:00:00 187 422 915 863 290 483 423 473
2018-01-01 09:00:00 270 144 749 710 983 755 839 709
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.