def handle_fitting_request(jobtype, fitting_task, model_name):
form = StartFittingForm()
mf = current_user.gc_metrics_file.first()
if mf is None:
flash('First upload your GoldenCheetah metrics file.')
return redirect(url_for('uploads.gc_metrics_file'))
metric_names = sorted(mf.csv_column_names())
metric_names = list(filter(lambda n: n not in metric_blacklist,
metric_names))
choices = list(zip(metric_names, metric_names))
form.load_metric_choice.choices = choices
form.perf_metric_choice.choices = choices
if form.validate_on_submit():
if current_user.has_pending_jobs(jobtype):
m = 'Sorry, this type of job is already pending for you.'
else:
if jobtype == JobType.ff_fitting:
algo = 'SLSQP'
else:
algo = 'LEAST_SQUARES'
load_metric = form.load_metric_choice.data
perf_metric = form.perf_metric_choice.data
fitting_task.delay(current_user.id, load_metric, perf_metric, algo)
'''perf_metric = '60 min Peak Power'
# perf_metric = '60m_critical_power'
# perf_metric = '60min Leistungsmaximum'
algos = ['LEAST_SQUARES', 'L-BFGS-B', 'TNC', 'SLSQP', 'CMA-ES',
'DE']
load_metrics = ['BikeScore', 'TSS', 'TRIMP Points',
'TRIMP Zonal Points', 'TRIMP(100) Points']
# load_metrics = ['skiba_bike_score', 'coggan_tss', 'trimp_points',
# 'trimp_zonal_points', 'trimp_100_points']
# load_metrics = ['BikeScore', 'TSS', 'TRIMP Punkte',
# 'TRIMP Punkte für Zonen', 'TRIMP(100) Punkte']
for algo in algos:
for load_metric in load_metrics:
t = fitting_task.delay(current_user.id,
load_metric,
perf_metric,
algo)
t.get()
print('algo {} for metric {} finished'.format(algo,
load_metric))
'''
m = Markup('Parameter fitting for {} has been started.<br>'
'Depending on server load, this can take a while.<br>'
'We\'ll mail you a notification when the job is done.'.
format(model_name))
flash(m)
return redirect(url_for('main.index'))
default_load_metric_idx = default_selection_index(mf.default_load_metric(),
metric_names)
default_perf_metric_idx = default_selection_index(mf.default_perf_metric(),
metric_names)
return render_template('training/fitting.html',
form=form,
model_name=model_name,
default_load_metric_index=default_load_metric_idx,
default_perf_metric_index=default_perf_metric_idx)
def pp_genplan():
form = GeneratePlanForm()
mf = current_user.gc_metrics_file.first()
if mf is None:
flash('First upload your GoldenCheetah metrics file.')
return redirect(url_for('uploads.gc_metrics_file'))
pp_parms = current_user.pp_parameters.first()
if pp_parms is None:
flash('First start your PerPot parameter fitting')
return redirect(url_for('training.pp_fitting'))
if current_user.is_over_pp_plans_limit():
flash(Markup('Sorry, you have reached the limit of {} PerPot plans.<br>'
'Delete some to generate new ones.'.
format(PLANS_LIMIT)))
return redirect(url_for('training.pp_myplans'))
if form.validate_on_submit():
if current_user.has_pending_jobs(JobType.pp_genplan):
m = 'Sorry, this type of job is already pending for you.'
else:
must_have_metrics = [pp_parms.load_metric, pp_parms.perf_metric]
start_date = next_monday_after(mf.last_training(must_have_metrics))
start_perf = current_user.pp_next_monday_pp()
off_weeks = sorted(parse_comma_separated_ints(form.off_weeks.data))
off_weeks = dec_list_elems(off_weeks) # users start at index 1
off_day_dates = parse_comma_separated_dates(form.off_days.data)
off_day_indexes = dates_to_indexes(start_date,
off_day_dates,
int(form.length.data))
goal = pp_parms.scale_perf_value(float(form.goal.data))
max_load = pp_parms.scale_load_value(float(form.max_load.data))
if form.min_load.data is None:
min_load = 0.0
else:
min_load = pp_parms.scale_load_value(float(form.min_load.data))
plan_request = PlanRequest(name=form.name.data.strip(),
model_parms=pp_parms,
start_date=start_date,
start_perf=start_perf,
goal=goal,
length=int(form.length.data),
max_load=max_load,
min_load=min_load,
off_weeks=off_weeks,
off_days=off_day_indexes,
weekly_cycle=parse_cycle_days(form))
pp_genplan_task.delay(current_user.id, plan_request)
# pp_genplan_cmaes_task.delay(current_user.id, plan_request)
m = Markup('PerPot plan generation has been started.<br>'
'Depending on server load, this can take a while.<br>'
'We\'ll mail you a notification when the job is done.')
flash(m)
return redirect(url_for('main.index'))
try:
next_monday_pp_scaled = current_user.pp_next_monday_pp()
print('next_monday_pp_scaled {}'.format(next_monday_pp_scaled))
next_monday_pp = pp_parms.unscale_perf_value(next_monday_pp_scaled)
print('next_monday_pp {}'.format(next_monday_pp))
must_have_metrics = [pp_parms.load_metric, pp_parms.perf_metric]
start_date = next_monday_after(mf.last_training(must_have_metrics))
form.max_load.label.text += \
' (e.g. 100 {})*'.format(pp_parms.load_metric)
return render_template('training/genplan.html',
form=form,
model='PerPot',
next_monday=start_date,
next_monday_p=next_monday_pp)
except:
m = Markup("Unfortunately your metrics export file can't be parsed.<br>"
"Please check if a needed column is missing or any values "
"are mangled.<br>"
"You can upload a new metrics export anytime.<br>"
"Please contact us, if you are sure your data is correct.")
flash(m)
return redirect(url_for('main.index'))
