def train_regressors(directory, assas, indices):
reg = {}
for assa in assas:
print assa
generic_path = directory + "matrices/cut_train_" + assa
data, val_ind = predict.load(generic_path)
val_ind = numpy.array(val_ind)
data = numpy.array(data)
# for i in range(len(data)-1) :
# a=len(data[i])
# b=len(data[i+1])
# if not a==b :
# print a
# print b
# print i
# print 'suivant'
# print data[0]
# print 'suivant'
# print data[1]
reg = predict.train(data, val_ind, indices[codecs.encode(assa, "utf_8")])
with open(directory + "test_reg_dict_" + assa + ".p", "w") as fp:
pickle.dump(reg, fp)
return reg
def train_regressors(directory,assas,indices):
reg = {}
for assa in assas :
print assa
generic_path = directory+'matrices/variant_train_'+assa
data,val_ind = predict.variant_load(generic_path)
val_ind=numpy.array(val_ind)
data = numpy.array(data)#[:,:10]
reg = predict.train(data,val_ind,indices[codecs.encode(assa,'utf_8')])
with open(directory+'rf_reg_dict_'+assa+'.p','w') as fp :
pickle.dump(reg,fp)
return reg
def train_command(args):
data, to_predict, true_data, tide_height_nans = process_data(normalise_data=True)
predictions, mean, var, _ = train(
to_predict,
data,
)
# filter the true tide_height to only be
# at the non_nan points
true_data_filtered = true_data.loc[tide_height_nans.values]
plot = GPPlot(
data,
true_data_filtered,
mean,
var,
[predictions],
TIDE_HEIGHT,
)
plot.init_plot()
plot.plot()
if args.save_figures:
plot.savefig(args.fig_name)
import predict
train_data_path, predict_data_path, predict_output_path = predict.download_data(
)
for model_id, model_type in predict.MODEL_CONFIGS:
predict.train(train_data_path, model_id, model_type, force_training=True)
general_df = load_dataset(dataset='general',
simfin_api_key=SIMFIN_API_KEY,
shareprices_df=shareprices_df)
banks_df = load_dataset(dataset='banks',
simfin_api_key=SIMFIN_API_KEY,
shareprices_df=shareprices_df)
insurance_df = load_dataset(dataset='insurance',
simfin_api_key=SIMFIN_API_KEY,
shareprices_df=shareprices_df)
# TRAIN
general_model = train(general_df,
winsor_quantile=0.01,
model_name='general_model',
feature_name='general',
param=dict(learning_rate=0.01,
max_depth=3,
subsample=.5,
colsample_bylevel=0.7,
colsample_bytree=0.7,
n_estimators=200))
banks_model = train(banks_df,
winsor_quantile=0.05,
model_name='banks_model',
feature_name='banks',
param=dict(learning_rate=0.01,
max_depth=2,
subsample=.8,
colsample_bylevel=0.7,
colsample_bytree=0.7,
n_estimators=200))