calculate_linear_cal_fun.py

# -*- coding: utf-8 -*-
"""
Main function used by Fig9_cal_cycle_calculation
"""

# headers
import numpy as np
import pandas as pd

# function to calculate the average and standard deviation at end of cal cycle
def calc_mean(gas_vals):
    rolling_mean = gas_vals.rolling(10).mean()
    rolling_std = gas_vals.rolling(10).std()
    mean_val = rolling_mean.iloc[-1] # last value
    std_val = rolling_std.iloc[-1]  # last value
    return mean_val, std_val

def calc_cal(COMA,ix):    
    # make DataFrame
    df_cal = pd.DataFrame({'time': COMA['time'][ix],
                           '[CO]d_ppb': COMA["[CO]d_ppm"][ix]*1000,
                           '[N2O]d_ppb': COMA["[N2O]d_ppm"][ix]*1000,
                           '[H2O]_ppm': COMA['[H2O]_ppm'][ix],
                           'AmbT_C': COMA['AmbT_C'][ix],
                           'Peak0': COMA['Peak0'][ix],
                           'AIN5': COMA['AIN5'][ix],
                           'AIN6': COMA['AIN6'][ix],
                           'GasP_torr': COMA['GasP_torr'][ix],
                           'SpectraID': COMA['SpectraID'][ix]
                           })
    
    # group data
    df_cal['groups'] = (df_cal.index.to_series().diff()>5).cumsum()
    df_grouped = df_cal.groupby('groups')
    
    # set up output variables
    feature_list = ['time','CO_val','CO_std','N2O_val','N2O_std',
                    'H2O','AmbT_C','Peak0','AIN5','AIN6',
                    'time_on']
    zero_data = np.zeros( shape = (len(df_grouped), len(feature_list)) )
    res = pd.DataFrame(zero_data, columns=feature_list)
    
    # look through cal cycles
    for ct, data in df_grouped:
        CO_series = pd.Series(data['[CO]d_ppb'].values) # make series for rolling attribute
        N2O_series = pd.Series(data['[N2O]d_ppb'].values) # make series for rolling attribute
        CellP_series = pd.Series(data['GasP_torr'].values)
        
        res.loc[ct,'time'] = data['time'].values[0] #start of sequence
        res.loc[ct,'CO_val'], res.loc[ct,'CO_std'] = calc_mean(CO_series)        
        res.loc[ct,'N2O_val'], res.loc[ct,'N2O_std'] = calc_mean(N2O_series)
        
        res.loc[ct,'AmbT_C'] = np.mean(data['AmbT_C'].values[-10:-1])
        res.loc[ct,'Peak0'] = np.mean(data['Peak0'].values[-10:-1])
        res.loc[ct,'AIN5'] = np.mean(data['AIN5'].values[-10:-1])
        res.loc[ct,'AIN6'] = np.mean(data['AIN6'].values[-10:-1])
        
        res.loc[ct,'GasP_val'], res.loc[ct,'GasP_std'] = calc_mean(CellP_series)
        
        res.loc[ct,'SpectraID'] = data['SpectraID'].values[-1] # last value
        res.loc[ct,'H2O'] = data['[H2O]_ppm'].values[-1] # last value
        
        #res.loc[ct,'start'] = min(data.index)
        #res.loc[ct,'end'] = max(data.index)
        
        #if ct == 0:
        #    res.loc[ct,'start_ct'] = 0
        #    res.loc[ct,'end_ct'] = len(data)-1
        #else:
        #    res.loc[ct,'start_ct'] = res.loc[ct-1,'end_ct'] + 1
        #    res.loc[ct,'end_ct'] = res.loc[ct,'start_ct'] + len(data)-1
            
    return res


# %% test curve fit
# Exponential fit to determine flush rate and steady state concentration
# (skip first several points where concentration seems to overshoot)
# probably difficult in general; need to look at cases one by one
# https://stackoverflow.com/questions/3938042/fitting-exponential-decay-with-no-initial-guessing

"""
from scipy.optimize import curve_fit
def func(x, a, b, c):
    return a*np.exp(-b*x) + c

x = np.linspace(3,60,58)
y = 200-data['CO_dry'].values[3:] # make curve descending

popt, pcov = curve_fit(func, x, y)

plt.plot(200-data['CO_dry'].values,'.')
plt.plot(x,func(x,popt[0],popt[1],popt[2]),'.')
"""