def D_parameters(dataset, D):
"""
Computes all 30 GA makespan solutions for each one
of the different threshold parameter values in D.
:returns: pd.dataframe (makespans), dictionary (exeqution times)
"""
# Create df to store makespan values
df = pd.DataFrame(columns=D)
# Initialize dictionary to store execution times
times = {}
for d in D: # iterate over D values
mk, time = solve_GA(dataset, D=d)
df.loc[:, d] = mk # Store makespans on new df column
times[d] = round(time, 3) # Round and store execution time
return df, times
def Pminit_parameters(dataset, Pminit):
"""
Computes all 30 GA makespan solutions for each one
of the different initial mutation probability values in Pminit.
:returns: pd.dataframe(makespans), dictionary (exeqution times)
"""
# Create df to store makespan values
df = pd.DataFrame(columns=Pminit)
# Initialize dictionary to store execution times
times = {}
for pminit in Pminit: # iterate over Pminit values
mk, time = solve_GA(dataset, Pminit=pminit)
df.loc[:, pminit] = mk # Store makespans on new df column
times[pminit] = round(time, 3) # Round and store execution time
return df, times
def Pc_parameters(dataset, Pc):
"""
Computes all 30 GA makespan solutions for each one
of the different crossover probability values in Pc.
:returns: pd.dataframe (makespans), dictionary (exeqution times)
"""
# Create df to store makespan values
df = pd.DataFrame(columns=Pc)
# Initialize dictionary to store execution times
times = {}
for pc in Pc: # iterate over Pc values
mk, time = solve_GA(dataset, Pc=pc)
df.loc[:, pc] = mk # Store makespans on new df column
times[pc] = round(time, 3) # Round and store execution time
return df, times
def M_parameters(dataset, M):
"""
This function takes as input a dataset with
job times on every machine and returns a
dataframe with all 30 GA makespan solutions
for each one of the different population
sizes, along with average times.
:returns: pd.dataframe (makespans), dictionary (exeqution times)
"""
# Create df to store makespan values for different
df = pd.DataFrame(columns=M)
# Initialize dictionary to store execution times
times = {}
for m in M: # iterate over population sizes
mk, time = solve_GA(dataset, M=m)
df.loc[:, m] = mk # Store makespans on new df column
times[m] = round(time, 3) # Round and store execution time
return df, times
ax3.set_ylabel(ylabel, fontfamily='serif')
# Make it more beautiful and save
plt.tight_layout(rect=[0, 0.03, 1, 0.92])
plt.savefig(car + '_all_parameters.pdf', facecolor=fig.get_facecolor())
# Use function to plot the effect
create4subplots(m_car6, pc_car6, pminit_car6, d_car6, titles[1], xlabels)
# After analyzing the plots, we obtain some optimal parameter values
# and we repeat GA on reC19 using these values
GA_reC19 = pd.DataFrame()
GA_reC19.loc[:, 'Default parameters'] = GA_df.loc[:, 'reC19']
GA_reC19.loc[:, 'Optimal parameters'], _ = solve_GA(reC19,
M=50,
Pc=0.9,
Pminit=0.6,
D=0.4)
# Visualize the default vs the optimal parameters
fig = plt.figure(facecolor='#FCF4DC')
red_square = dict(markerfacecolor='r', marker='s', markersize=4)
GA_reC19.plot(kind='box',
patch_artist=True,
flierprops=red_square,
medianprops={
'linestyle': '-',
'linewidth': 5
},
showmeans=True,
color={'medians': 'red'})