# TSNE original | TSNE original early-stop | TSNE chain early-stop | TSNE chain
# Each row is the embeddings for each perplexity
import os
import flask
import pandas as pd
from flask import render_template
from string import Template
from common.dataset import dataset
from icommon import hyper_params
dir_path = os.path.dirname(os.path.realpath(__file__))
data_dir = f"{dir_path}/data"
dataset.set_data_home(data_dir)
app = flask.Flask("render_view_app")
CSV_PATH = "./plot_chain"
TABLE = """
<table id="${tbl_id}">
<thead> ${thead} </thead>
<tbody> ${tbody} </tbody>
</table>
"""
FIG_IMG = """
<img id='img_${idx}' class='img-thumbnail' alt='xxx'
def run_with_sklearn(data, labels):
for model in [PCA, TSNE]:
Z = model(n_components=2).fit_transform(data)
metric = DRMetric(data, Z)
auc_rnx = metric.auc_rnx()
fig = get_fig_plot_z2d(
Z, title=f"sklearn/{model.__name__}, auc_rnx={auc_rnx:.3f}", with_imgs=True
)
fig.savefig(f"./plots/{args.dataset_name}_sklearn_{model.__name__}.png")
plt.close(fig)
if __name__ == "__main__":
dataset.set_data_home("./data")
help_msg = """
Run DeepPPCAModel with custom params:
$ python ppca_model2.py -d "DIGITS" -hd 50 -lr 0.0075 -n 2000
"""
ap = argparse.ArgumentParser(description=help_msg)
ap.add_argument("-r", "--run_id", default="999")
ap.add_argument("-d", "--dataset_name", default="")
ap.add_argument("-x", "--dev", action="store_true")
ap.add_argument("-lr", "--learning_rate", default=1e-3, type=float)
ap.add_argument("-s", "--scale_data", default="unitScale")
ap.add_argument("-n", "--n_iters", default=1000, type=int)
ap.add_argument(
"-hd",
"--hidden_dim",
mlflow.log_param('n_iters', n_iters)
mlflow.log_param('lr', lr)
plot_args = {
'prefix': f'{dataset_name}-{lr}-{n_iters}',
'suffix': '.png',
'save_to_file': SAVE_FIGURES,
'track_with_mlflow': TRACK_FLOW,
}
X_original, X, y = dataset.load_dataset(dataset_name)
run_ppca(X_original, X, y, lr, n_iters, plot_args)
if __name__ == '__main__':
common.plot.simple_plot.PLOT_DIR = './plots'
dataset.set_data_home('./data')
SAVE_FIGURES = True
TRACK_FLOW = True
# mlflow.set_experiment('scikit-learn')
# mlflow.log_param('dataset_name', dataset_name)
# run_with_sklearn(data, original_data=X, labels_true=y)
mlflow.set_experiment('PPCA_pyro')
learning_rates = [
0.005, 0.01, 0.02, 0.025, 0.005, 0.075, 0.1, 0.15, 0.2, 0.5
]
n_iters = 250
datasets = ['COUNTRY_2014']
def get_data_with_all_labels(dataset_name):
dataset.set_data_home("./data")
return dataset.load_dataset_multi_label(dataset_name)
if __name__ == "__main__":
import argparse
ap = argparse.ArgumentParser()
ap.add_argument("-d", "--dataset_name", default="")
ap.add_argument("-n",
"--n_links",
default=0,
type=int,
help="# of links each type")
ap.add_argument("-ns", "--n_sim_links", default=None, type=int)
ap.add_argument("-nd", "--n_dis_links", default=None, type=int)
ap.add_argument("-s", "--seed", default=2019, type=int)
ap.add_argument("-x", "--dev", action="store_true")
args = ap.parse_args()
dir_path = os.path.dirname(os.path.realpath(__file__))
LINK_DIR = f"{dir_path}/links"
DATA_DIR = f"{dir_path}/data"
dataset.set_data_home(DATA_DIR)
# always makesure to set random seed for reproducing
random.seed(args.seed)
n_sim = args.n_sim_links or args.n_links
n_dis = args.n_dis_links or args.n_links
gen_constraints(args.dataset_name, n_sim, n_dis)
if args.dev:
test_auto_generated_constraints(args.dataset_name, n_sim, n_dis)
