simtk.org

Code: Select all

import pandas as pd

Code: Select all

dfCMC_q = pd.read_csv(r"CMC\Kinematics_q.sto", header=8, sep="\t", index_col="time")

Code: Select all

def sync_dfs(df, df1, rolling_mean_size=3):
    """
    From two data frames sampled at slightly different times, return two dataframes 
    at synchronized time points. It uses linear interpolation to fill missing data, 
    and returns an averaged over some time points data.
    
    Example: 
        df has 10 entries, each at half a second (.5, 1.5, 2.5 ...)
        df1 has 30 entries, each at every third of a second (.33, .66, 1, 1.33, ...)
        rolling_mean_size = 3
        
        this function first computes a dataframe "whole" containing time points (.33, .5, .66, 1, 1.33, 1.5, ...)
        it fills the missing values by linearly interpolating between two know data points (e.g. between .5 and 1.5)
        it then averages over 3 time points, resulting in a "whole" data frame of size 40/3
        
        And it finally returns two dataframes, adaptations of the input ones, both of the same size and sampling times.
    """
    whole = pd.merge(df.reset_index(), df1.reset_index(), how='outer', on='time', suffixes=('_x', '_y')).set_index("time").sort_index().interpolate()
    whole = whole.rolling(rolling_mean_size).mean()
    df_ = whole.loc[:, whole.columns.str.endswith('x')].rename(columns = lambda x : str(x).rsplit('_', 1)[0])
    df1_ = whole.loc[:, whole.columns.str.endswith('y')].rename(columns = lambda x : str(x).rsplit('_', 1)[0])
    return df_, df1_