fr1ll · October 6, 2020 02:41
diff --git a/2020-10-05_timing-test-map-replace.py b/2020-10-05_timing-test-map-replace.py
 import pandas as pd
 import numpy as np
 import timeit

 def random_series(count=10000):
    vals = np.random.randint(0,20*count,count)
    return pd.Series(data=vals, name="RANDVALS")

 def unique_map(s: pd.Series):
    as_unique = s.unique()
    return {dval: i for i, dval in enumerate(as_unique)}
    
 funcs = ["s.replace(d_map)",
         "s.map(d_map)",
         "s.map(d_map.get)"]  
 counts = [10, 100, 1000, 10000, 100000]

 def time_some_counts(func, counts):
    times = []
    for c in counts:
        # create data with lenght set by c
        s = random_series(c)
        d_map = unique_map(s)
        
        # tricky way to pass data to timeit
        myGlobals = globals()
        myGlobals.update({'s': s, 'd_map': d_map})
        
        #time the functions
        new_times = timeit.Timer(func, globals=myGlobals).autorange()
        
        
        new_times = [func, c, new_times[0], new_times[1], new_times[1]/new_times[0]]
        times.append(new_times)
    return times
  
 columns = ["func", "length_of_series", "runs", "total_runtime", "mean_runtime"]
 df = pd.DataFrame(columns=columns)

 for f in funcs:
    times = time_some_counts(f, counts)
    df = df.append(pd.DataFrame(times, columns=columns))
    
    
 import altair as alt

 chart = alt.Chart(df).mark_circle(size=60).encode(
    x=alt.X('length_of_series', scale=alt.Scale(type="log")),
    y=alt.Y('mean_runtime', scale=alt.Scale(type="log")),
    color='func',
 )

 chart + chart.transform_regression('length_of_series', 'mean_runtime', groupby=["func"], method="pow").mark_line()
No results found