Change index at DataFrame creation in get_agent_vars_dataframe

[Tortar]

This is slightly faster as one would expect, code to reproduce:

import timeit
a = timeit.timeit("s =pd.DataFrame.from_records(t); s = s.set_index(['a', 'b'])","import pandas as pd\nq=list(range(1000)); t={'a':q,'b':q}", number=1000)
b = timeit.timeit("s =pd.DataFrame.from_records(t, index=['a', 'b'])","import pandas as pd\nq=list(range(1000)); t={'a':q,'b':q}", number=1000)
print(a, b)

[codecov]

Codecov Report

Base: 82.07% // Head: 82.06% // Decreases project coverage by -0.02% ⚠️

Coverage data is based on head (dcbc90b) compared to base (c7b2ccf).
Patch has no changes to coverable lines.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1586      +/-   ##
==========================================
- Coverage   82.07%   82.06%   -0.02%     
==========================================
  Files          18       18              
  Lines        1389     1388       -1     
  Branches      271      271              
==========================================
- Hits         1140     1139       -1     
  Misses        205      205              
  Partials       44       44              

Impacted Files	Coverage Δ
mesa/datacollection.py	95.29% <ø> (-0.06%)	⬇️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[rht]

This is slightly faster ...

By how much?

[Tortar]

25% in this case, expected more, but it seems that indexing costs also at creation, but more you have columns less it impacts

[Tortar]

Actually it is almost the same with more columns

import timeit
a = timeit.timeit("s =pd.DataFrame.from_records(t); s = s.set_index(['a', 'b'])","import pandas as pd\nq=list(range(10000)); t={x:q for x in 'abcdefghilmn'}", number=100)
b = timeit.timeit("s =pd.DataFrame.from_records(t, index=['a', 'b'])","import pandas as pd\nq=list(range(10000)); t={x:q for x in 'abcdefghilmn'}", number=100)
print(a, b)

gives me anyway a 15% better where we have 12 columns

import timeit

a = timeit.timeit("s =pd.DataFrame.from_records(t); s = s.set_index(['a', 'b'])","import pandas as pd\nq=list(range(10000)); cols = 'abcdefghilmnopqrstuvz123456'; t={x:q for x in cols}", number=100)
b = timeit.timeit("s =pd.DataFrame.from_records(t, index=['a', 'b'])","import pandas as pd\nq=list(range(10000)); cols = 'abcdefghilmnopqrstuvz123456'; t={x:q for x in cols}", number=100)
print(a, b)

gives me anyway a 30% better where we have 30+ columns

[rht]

LGTM