Spaces:
Sleeping
Sleeping
File size: 44,307 Bytes
2875fe6 a785d5a aac28f7 a785d5a ffdaaba a785d5a 2875fe6 a785d5a 2875fe6 a785d5a ffdaaba a785d5a aac28f7 2875fe6 aac28f7 2875fe6 a785d5a 2875fe6 a785d5a aac28f7 a785d5a aac28f7 a785d5a 2875fe6 a785d5a 2875fe6 a785d5a 2875fe6 a785d5a ffdaaba a785d5a 2875fe6 a785d5a 2875fe6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 |
import gradio as gr
import pandas as pd
import plotly.graph_objects as go
import numpy as np
import torch
from typing import Dict, List, Tuple
import re
from typing import Callable, Union, Dict
class TimeSeriesEditor:
def __init__(self, seq_length: int, feature_dim: int, trainer):
# Existing initialization
self.seq_length = seq_length
self.feature_dim = feature_dim
self.trainer = trainer
self.coef = None
self.stepsize = None
self.sampling_steps = None
self.feature_names = ["revenue", "download", "daily active user"]# * 20
# self.feature_names = [f"Feature {i}" for i in range(self.feature_dim)]
# Store the latest model output
self.latest_sample = None
self.latest_observed_points = None
self.latest_observed_mask = None
self.latest_gradient_control_signal = None
self.latest_model_control_signal = None
# self.latest_metrics
# Define scales for each feature
self.feature_scales = {
0: 1000000, # Revenue: $1M per 0.1
1: 100000, # Download: 100K downloads per 0.1
2: 10000 # AU: 10K active users per 0.1
}
self.feature_units = {
0: "$", # Revenue
1: "downloads", # Download
2: "users" # AU
}
self.show_normalized = True
# Add frequency band multipliers
self.freq_bands = np.ones(5) # 5 frequency bands, initially all set to 1.0
self.function_parser = FunctionParser()
self.trending_controls = [
# (200, 250, 0, self.function_parser.string_to_function("sin(2*pi*x)"), 0.05)
# 200,250,0,sin(2*pi*x),0.05
]
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
def format_value(self, value: float, feature_idx: int) -> str:
"""Format value with appropriate units and notation"""
if self.show_normalized:
return f"{value:.4f}"
else:
if feature_idx == 0: # Revenue
return f"{self.feature_units[feature_idx]}{value:,.2f}"
else: # Downloads and AU
return f"{value:,.0f} {self.feature_units[feature_idx]}"
def create_plot(self, sample: np.ndarray, observed_points: torch.Tensor,
observed_mask: torch.Tensor,
gradient_control_signal: Dict, metrics: Dict) -> List[go.Figure]:
figures = []
# Get weights from model_control_signal (will be all 1s if not provided)
weights = observed_mask
for feat_idx in range(self.feature_dim):
fig = go.Figure()
# Scale values if needed
scale_factor = self.feature_scales[feat_idx] * 10 if not self.show_normalized else 1
# Plot predicted line
predicted_values = sample[:, feat_idx] * scale_factor
fig.add_trace(go.Scatter(
x=np.arange(self.seq_length),
y=predicted_values,
mode='lines',
name='Predicted',
line=dict(color='green', width=2),
showlegend=True
))
# Calculate and plot confidence bands based on weights
# Lower weights = larger uncertainty bands
mask = observed_points[:, feat_idx] > 0
ox = np.arange(0, self.seq_length)[mask]
oy = observed_points[mask, feat_idx].numpy() * scale_factor
weights_masked = 1 - weights[mask, feat_idx].numpy()
# Calculate error bars - inverse relationship with weight
# Weight of 1.0 gives minimal uncertainty (0.02)
# Weight of 0.1 gives larger uncertainty (0.2)
# error_y = 0.02 / weights_masked
error_y = weights_masked / 5
# Plot observed points with error bars - changed symbol to 'cross'
fig.add_trace(go.Scatter(
x=ox,
y=oy,
mode='markers',
name='Observed',
marker=dict(
# special red
color='rgba(255, 0, 0, 0.5)',
# size=10,
symbol='x', # Changed from 'circle' to 'x' for cross symbol
),
error_y=dict(
type='data',
array=error_y * scale_factor,
visible=True,
thickness=0.5,
width=2,
color='blue'
),
showlegend=True
))
# Add shaded confidence bands around the predicted line
# This shows the general uncertainty in the prediction
uncertainty = 0.05 # Base uncertainty level
upper_bound = predicted_values + uncertainty * scale_factor
lower_bound = predicted_values - uncertainty * scale_factor
fig.add_trace(go.Scatter(
x=np.concatenate([np.arange(self.seq_length), np.arange(self.seq_length)[::-1]]),
y=np.concatenate([upper_bound, lower_bound[::-1]]),
# fill='toself',
# fillcolor='rgba(0,100,0,0.1)',
line=dict(color='rgba(255,255,255,0)'),
name='Prediction Interval',
showlegend=True
))
# Add vertical lines for peak points
if gradient_control_signal.get("peak_points"):
for peak_point in gradient_control_signal["peak_points"]:
fig.add_vline(x=peak_point, line_dash="dash", line_color="red")
# Add metrics annotations
total_value = np.sum(sample[:, feat_idx]) * (self.feature_scales[feat_idx] * 10 if not self.show_normalized else 1)
annotations = [dict(
x=0.02,
y=1.1,
xref="paper",
yref="paper",
text=f"Total {self.feature_names[feat_idx]}: {self.format_value(total_value, feat_idx)}",
showarrow=False
)]
# Update y-axis title based on feature and scaling
if self.show_normalized:
y_title = f'{self.feature_names[feat_idx]} (Normalized)'
else:
unit = self.feature_units[feat_idx]
y_title = f'{self.feature_names[feat_idx]} ({unit})'
# Create a more informative legend for uncertainty
legend_text = (
"Prediction with Confidence Bands<br>"
"• Blue points: Observed values with uncertainty<br>"
"• Green line: Predicted values<br>"
# "• Shaded area: Prediction uncertainty<br>"
"• Error bars: Observation uncertainty (larger = lower weight)"
)
fig.update_layout(
title=dict(
text=f'Feature: {self.feature_names[feat_idx]}',
x=0.5,
y=0.95
),
xaxis_title='Time',
yaxis_title=y_title,
height=400,
showlegend=True,
dragmode='select',
annotations=[
*annotations,
# dict(
# x=1.15,
# y=0.5,
# xref="paper",
# yref="paper",
# text=legend_text,
# showarrow=False,
# align="left",
# bordercolor="black",
# borderwidth=1,
# borderpad=4,
# bgcolor="white",
# )
],
margin=dict(r=200) # Add right margin for legend
)
figures.append(fig)
return figures
def update_scaling(self,
revenue_scale: float,
download_scale: float,
au_scale: float,
show_normalized: bool) -> Tuple[List[go.Figure], Dict]:
"""Update the scaling parameters and redraw plots"""
if self.latest_sample is None:
return [], {}
# Update scales
self.feature_scales = {
0: revenue_scale,
1: download_scale,
2: au_scale
}
self.show_normalized = show_normalized
# Calculate metrics
metrics = {
'show_normalized': self.show_normalized
}
for feat_idx in range(self.feature_dim):
total = np.sum(self.latest_sample[:, feat_idx]) * (self.feature_scales[feat_idx] * 10 if not self.show_normalized else 1)
metrics[f'total_{self.feature_names[feat_idx]}'] = self.format_value(total, feat_idx)
# Update plots
figures = self.create_plot(
self.latest_sample,
self.latest_observed_points,
self.latest_observed_mask,
self.latest_gradient_control_signal,
metrics
)
return figures, metrics
def parse_data_points(self, df) -> Dict:
"""Parse data points from DataFrame with columns: time,feature,value"""
data_dict = {}
if df is None or df.empty:
return data_dict
for _, row in df.iterrows():
# Skip if any required value is NaN
if pd.isna(row['time']) or pd.isna(row['feature']) or pd.isna(row['value']):
continue
try:
time_idx = int(row['time'])
feature_idx = int(row['feature'])
value = float(row['value'])
if time_idx not in data_dict:
data_dict[time_idx] = {}
data_dict[time_idx][feature_idx] = (value, 1.0)
except (ValueError, TypeError):
continue
return data_dict
def parse_point_groups(self, df) -> Dict:
"""Parse point groups from DataFrame with columns: start,end,interval,feature,value,weight"""
data_dict = {}
if df is None or df.empty:
return data_dict
for _, row in df.iterrows():
# Skip if any required value is NaN
if pd.isna(row['start']) or pd.isna(row['end']) or pd.isna(row['interval']) or \
pd.isna(row['feature']) or pd.isna(row['value']):
continue
try:
start = int(row['start'])
end = int(row['end'])
interval = int(row['interval'])
feature = int(row['feature'])
value = float(row['value'])
weight = float(row.get('weight', 1.0)) if not pd.isna(row.get('weight')) else 1.0
for t in range(start, end + 1, interval):
if 0 <= t < self.seq_length:
if t not in data_dict:
data_dict[t] = {}
data_dict[t][feature] = (value, weight)
except (ValueError, TypeError):
continue
return data_dict
def to_tensor(self, observed_points_dict, seq_length, feature_dim):
observed_points = torch.zeros((seq_length, feature_dim))
observed_weights = torch.zeros((seq_length, feature_dim))
for seq, feature_dict in observed_points_dict.items():
for feature, (value, weight) in feature_dict.items():
observed_points[seq, feature] = value
observed_weights[seq, feature] = weight
return observed_points, observed_weights
def apply_direct_edits(self, sample: np.ndarray, edit_params: Dict) -> np.ndarray:
"""Apply direct edits to the sample array"""
edited_sample = sample.copy()
if edit_params.get("enable_direct_area"):
areas = self.parse_area_selections(edit_params["direct_areas"])
for area in areas:
start, end, feat_idx, target = area
edited_sample[start:end, feat_idx] += target
edited_sample = np.clip(edited_sample, 0, 1)
return edited_sample
def parse_area_selections(self, area_text: str) -> List[Tuple]:
"""Parse area selection text into (start, end, feature, target) tuples"""
areas = []
if not area_text.strip():
return areas
area_text = area_text.replace('\n', ';')
for line in area_text.strip().split(';'):
if not line.strip():
continue
try:
start, end, feat, target = map(float, line.strip().split(','))
areas.append((int(start), int(end), int(feat), target))
except (ValueError, IndexError):
continue
return areas
def apply_trending_mask(self, points: torch.Tensor, mask: torch.Tensor, consider_last_generated=False) -> Tuple[torch.Tensor, torch.Tensor]:
"""Apply trending functions as soft constraints through masks"""
if not self.trending_controls or self.latest_sample is None:
return points, mask
for start, end, feat_idx, func, confidence in self.trending_controls:
if start < 0 or end > self.seq_length or start >= end:
continue
# Generate x values normalized between 0 and 1 for the segment
x = np.linspace(0, 1, end - start)
try:
# Calculate the function values
y = func(x)
# Scale the function output to 0-1 range
y = (y - np.min(y)) / (np.max(y) - np.min(y))
# points[start:end, feat_idx] = torch.tensor(y, dtype=points.dtype)
# mask[start:end, feat_idx] = max(mask[start:end, feat_idx], min(1.0, confidence * abs(
# self.latest_sample[start:end, feat_idx] - y
# ))) # Use lower weight for trending constraints
except Exception as e:
print(f"Error applying function: {e}")
continue
# Apply the trend as soft constraints
mask_zero = (mask[start:end, feat_idx] == 0)
points[start:end, feat_idx][mask_zero] = torch.tensor(y, dtype=points.dtype)[mask_zero]
mask[start:end, feat_idx][mask_zero] = torch.tensor(confidence * np.ones_like(y), dtype=mask.dtype)[mask_zero]
# mask[start:end, feat_idx][mask_zero] = torch.tensor((confidence * np.abs(self.latest_sample[start:end, feat_idx] - y)), dtype=mask.dtype)[mask_zero]
mask = mask.clamp(0, 1)
return points, mask
def update_model(self,
figures: List[go.Figure],
data_points: str,
point_groups: str,
enable_area_control: bool,
area_selections: str,
enable_auc: bool,
auc_value: float,
enable_peaks: bool,
peak_points: str,
peak_alpha: float,
auc_weight: float,
peak_weight: float,
enable_trending: bool = True,
enable_trending_with_diff: bool = False,
trending_params: str = ""
) -> Tuple[List[go.Figure], str, str, Dict]:
# Parse both point groups and individual data points
individual_points_dict = self.parse_data_points(data_points)
group_points_dict = self.parse_point_groups(point_groups)
# Merge dictionaries, giving preference to individual points
combined_points_dict = group_points_dict.copy()
for t, feat_dict in individual_points_dict.items():
if t not in combined_points_dict:
combined_points_dict[t] = {}
for f, v in feat_dict.items():
combined_points_dict[t][f] = v
# Convert to tensor
observed_points, observed_weights = self.to_tensor(
combined_points_dict,
self.seq_length,
self.feature_dim
)
observed_mask = observed_weights
# Parse peak points
peak_points_list = []
if enable_peaks and peak_points:
try:
peak_points_list = [int(x.strip()) for x in peak_points.split(',') if x.strip()]
except ValueError:
peak_points_list = []
# Apply trending control if enabled
if enable_trending and trending_params:
self.parse_trending_parameters(trending_params)
observed_points, observed_mask = self.apply_trending_mask(observed_points, observed_mask, consider_last_generated=enable_trending_with_diff)
# Build gradient control signal
# IMPORTANT
gradient_control_signal = {}
if enable_auc:
gradient_control_signal["auc"] = auc_value
gradient_control_signal["auc_weight"] = auc_weight
if enable_peaks:
gradient_control_signal.update({
"peak_points": peak_points_list,
"peak_alpha": peak_alpha,
"peak_weight": peak_weight
})
# Build model control signal
model_control_signal = {}
# if enable_area_control and area_selections:
# areas = self.parse_area_selections(area_selections)
# if areas:
# model_control_signal["selected_areas"] = areas
# Run prediction
with torch.no_grad():
# to cuda
observed_points = observed_points.to(self.device)
observed_mask = observed_mask.to(self.device)
sample = self.trainer.predict_weighted_points(
observed_points, # (seq_length, feature_dim)
observed_mask, # (seq_length, feature_dim)
self.coef, # fixed
self.stepsize, # fixed
self.sampling_steps, # fixed
# model_control_signal=model_control_signal,
gradient_control_signal=gradient_control_signal
)
observed_points = observed_points.cpu()
observed_mask = observed_mask.cpu()
# Store latest results
self.latest_sample = sample
self.latest_observed_points = observed_points
self.latest_observed_mask = observed_mask
self.latest_gradient_control_signal = gradient_control_signal
self.latest_model_control_signal = model_control_signal
# Calculate metrics
metrics = {
'show_normalized': self.show_normalized
}
for feat_idx in range(self.feature_dim):
total = np.sum(sample[:, feat_idx]) * (self.feature_scales[feat_idx] * 10 if not self.show_normalized else 1)
metrics[f'total_{self.feature_names[feat_idx]}'] = self.format_value(total, feat_idx)
# Update plots
figures = self.create_plot(sample, observed_points, observed_mask, gradient_control_signal, metrics)
return figures, data_points, point_groups, metrics
def update_additional_edit(
self,
enable_direct_area: bool,
direct_areas: str):
# Apply direct edits if enabled
if enable_direct_area:
sample = self.apply_direct_edits(self.latest_sample, {
"enable_direct_area": enable_direct_area,
"direct_areas": direct_areas
})
else:
sample = self.latest_sample
# Calculate metrics
metrics = {
'show_normalized': self.show_normalized
}
for feat_idx in range(self.feature_dim):
total = np.sum(sample[:, feat_idx]) * (self.feature_scales[feat_idx] * 10 if not self.show_normalized else 1)
metrics[f'total_{self.feature_names[feat_idx]}'] = self.format_value(total, feat_idx)
# Update plots
figures = self.create_plot(
sample,
self.latest_observed_points,
self.latest_observed_mask,
self.latest_gradient_control_signal,
metrics
)
return figures, metrics
def apply_frequency_filter(self, signal: np.ndarray) -> np.ndarray:
"""Apply FFT-based frequency filtering using the current band multipliers"""
# Get FFT of the signal
fft = np.fft.fft(signal)
freqs = np.fft.fftfreq(len(signal))
# Split frequencies into 5 bands
# Exclude DC component (0 frequency) from bands
pos_freqs = freqs[1:len(freqs)//2]
freq_ranges = np.array_split(pos_freqs, 5)
# Apply band multipliers
filtered_fft = fft.copy()
# Handle DC component separately (lowest frequency)
filtered_fft[0] *= self.freq_bands[4] # Apply very low freq multiplier to DC
# Apply multipliers to each frequency band
for i, freq_range in enumerate(freq_ranges):
# Get indices for this frequency band
band_mask = np.logical_and(
freqs >= freq_range[0],
freqs <= freq_range[-1]
)
# Apply multiplier to positive and negative frequencies symmetrically
filtered_fft[band_mask] *= self.freq_bands[4-i]
filtered_fft[np.flip(band_mask)] *= self.freq_bands[4-i]
# Convert back to time domain
filtered_signal = np.real(np.fft.ifft(filtered_fft))
return filtered_signal
def update_frequency_bands(self, band_idx: int, value: float) -> Tuple[List[go.Figure], Dict]:
"""Update a frequency band multiplier and recompute the filtered signal"""
if self.latest_sample is None:
return [], {}
# Update the specified band multiplier
self.freq_bands[band_idx] = value
# Apply frequency filtering to each feature
filtered_sample = self.latest_sample.copy()
for feat_idx in range(self.feature_dim):
filtered_sample[:, feat_idx] = self.apply_frequency_filter(
self.latest_sample[:, feat_idx]
)
# Ensure values remain in valid range
filtered_sample = np.clip(filtered_sample, 0, 1)
# Calculate metrics
metrics = {
'show_normalized': self.show_normalized,
'frequency_bands': self.freq_bands.tolist()
}
for feat_idx in range(self.feature_dim):
total = np.sum(filtered_sample[:, feat_idx]) * (self.feature_scales[feat_idx] * 10 if not self.show_normalized else 1)
metrics[f'total_{self.feature_names[feat_idx]}'] = self.format_value(total, feat_idx)
# Update plots
figures = self.create_plot(
filtered_sample,
self.latest_observed_points,
self.latest_observed_mask,
self.latest_gradient_control_signal,
metrics
)
return figures, metrics
def parse_trending_parameters(self, trending_text: str) -> List[Tuple]:
"""Parse trending control parameters into (start, end, feature, function) tuples"""
trending_params = []
if not trending_text.strip():
return trending_params
trending_text = trending_text.replace('\n', ';')
for line in trending_text.strip().split(';'):
if not line.strip():
continue
try:
# Split by comma and handle the function part separately
parts = line.strip().split(',', 4)
if len(parts) != 5:
continue
start, end, feat = map(int, parts[:3])
function_str = parts[3].strip()
confidence = float(parts[4])
# Convert the function string to a callable
try:
func = self.function_parser.string_to_function(function_str)
trending_params.append((start, end, feat, func, confidence))
except ValueError as e:
print(f"Error parsing function '{function_str}': {e}")
continue
except (ValueError, IndexError):
continue
self.trending_controls = trending_params # Store the parsed parameters
return trending_params
def create_gradio_interface(editor: TimeSeriesEditor):
with gr.Blocks() as app:
gr.Markdown("# Time Series Editor")
gr.Markdown("## Instruction: Scroll Down + Click [Update Figure] [~10s]")
metrics_display = gr.JSON(label="Metrics", value={})
with gr.Row():
with gr.Column(scale=1):
# with Tab():
# Scaling Parameters Section
# with gr.Group():
gr.Markdown("## Scaling Parameters")
with gr.Accordion("Open for More Detail", open=False):
revenue_scale = gr.Number(
label="Revenue Scale ($ per 0.1 in model)",
value=1000000
)
download_scale = gr.Number(
label="Download Scale (downloads per 0.1 in model)",
value=100000
)
au_scale = gr.Number(
label="Active Users Scale (users per 0.1 in model)",
value=10000
)
show_normalized = gr.Checkbox(
label="Show Normalized Values (0-1 scale)",
value=True
)
update_scaling_btn = gr.Button("Update Scaling")
# TS Section
gr.Markdown("## Time Series Control Panel")
# with gr.Accordion("Open for More Detail"):
with gr.Group():
gr.Markdown("### Fixed Point Control")
data_points_df = gr.Dataframe(
headers=["time", "feature", "value"],
datatype=["number", "number", "number"],
# label="Anchor Point Control",
value=[[0, 0, 0.04], [2, 0, 0.58], [6, 0, 0.27], [58, 0, 0.8], [60, 0, 0.5]],
col_count=(3, "fixed"), # Fix number of columns
interactive=True
)
add_data_point_btn = gr.Button("Add Data Point")
def add_data_point(df):
new_row = pd.DataFrame([[None, 0, None]],
columns=["time", "feature", "value"])
return pd.concat([df, new_row], ignore_index=True)
add_data_point_btn.click(
fn=add_data_point,
inputs=[data_points_df],
outputs=[data_points_df]
)
with gr.Group():
gr.Markdown("### Group of Anchor Point Control with Confidence")
point_groups_df = gr.Dataframe(
headers=["start", "end", "interval", "feature", "value", "weight"],
datatype=["number", "number", "number", "number", "number", "number"],
# label="Group of Anchor Point Control",
value=[[0, 50, 10, 0, 0.5, 0.1], [100, 150, 50, 0, 0.1, 0.5]],
col_count=(6, "fixed"), # Fix number of columns
interactive=True
)
add_point_group_btn = gr.Button("Add Point Group")
def add_point_group(df):
new_row = pd.DataFrame([[None, None, None, 0, None, None]],
columns=["start", "end", "interval", "feature", "value", "weight"])
return pd.concat([df, new_row], ignore_index=True)
add_point_group_btn.click(
fn=add_point_group,
inputs=[point_groups_df],
outputs=[point_groups_df]
)
with gr.Group():
# with gr.Tab("Trending Control"):
gr.Markdown("### Trending Control")
gr.Markdown("""
Enter trending control parameters in the format:
```
start_time,end_time,feature,function,confidence
```
Examples:
- Linear trend: `0,100,0,x`
- Sine wave: `0,100,0,sin(2*pi*x)`
- Exponential: `0,100,0,exp(-x)`
Separate multiple trends with semicolons.
""")
enable_trending_control = gr.Checkbox(label="Enable Trending Control", value=False)
enable_trending_control_with_diff = gr.Checkbox(label="Consider Last Generated", value=False)
trending_control = gr.Textbox(
label="Trending Control Parameters",
lines=2,
placeholder="Enter parameters: start_time,end_time,feature,function,condifdence; separated by semicolons",
value="200,250,0,sin(2*pi*x),0.05"
)
# Area Control Parameters
with gr.Group(visible=False):
gr.Markdown("### Area Control")
enable_area_control = gr.Checkbox(label="Enable Area Control", value=False)
area_selections = gr.Textbox(
label="Area Selections (format: start_time,end_time,feature,target_value)",
lines=2,
placeholder="Enter areas: start,end,feature,target; separated by semicolons",
)
# AUC Parameters
gr.Markdown("### Statistics Control")
enable_auc = gr.Checkbox(label="Enable Total Sum Control", value=True)
auc_input = gr.Number(label="Target Sum Value", value=-150)
auc_weight_input = gr.Number(label="Sum Weight", value=10.0)
# Peak Parameters
with gr.Group(visible=False):
gr.Markdown("### Peak Control")
enable_peaks = gr.Checkbox(label="Enable Peak Control", value=False)
peak_points_input = gr.Textbox(label="Peak Points (comma-separated)", value="100,200")
peak_alpha_input = gr.Number(label="Peak Alpha", value=10)
peak_weight_input = gr.Number(label="Peak Weight", value=1.0)
update_model_btn = gr.Button("Update Figure")
gr.Markdown("## Extend Edit", visible=False)
with gr.Tab("Range Shift", visible=False):
# gr.Markdown("### Direct Edit Control")
enable_direct_area = gr.Checkbox(label="Enable Direct Edits", value=False) # range shift
direct_areas = gr.Textbox(
label="Direct Edit Areas (format: start_time,end_time,feature,delta)",
lines=2,
placeholder="Enter areas: start,end,feature,delta; separated by semicolons",
value="150,200,0,-0.1"
)
update_additional_btn = gr.Button("Update Additional Edit")
# with gr.Tab("Trending Control"):
# gr.Markdown("### Trending Control")
# gr.Markdown("""
# Enter trending control parameters in the format:
# ```
# start_time,end_time,feature,function
# ```
# Examples:
# - Linear trend: `0,100,0,x`
# - Sine wave: `0,100,0,sin(2*pi*x)`
# - Exponential: `0,100,0,exp(-x)`
# Separate multiple trends with semicolons.
# """)
# enable_trending_control = gr.Checkbox(label="Enable Trending Control", value=False)
# enable_trending_control_with_diff = gr.Checkbox(label="Consider Last Generated", value=False)
# trending_control = gr.Textbox(
# label="Trending Control Parameters",
# lines=2,
# placeholder="Enter parameters: start_time,end_time,feature,function,condifdence; separated by semicolons",
# value="0,100,0,sin(2*pi*x),0.3"
# )
# with gr.Tab("Frequency Controls", visible=False):
with gr.Group(visible=False):
gr.Markdown("Adjust multipliers for different frequency bands (0-2)")
freq_bands = [
gr.Slider(
minimum=0, maximum=2, step=0.1, value=1.0,
label=f"Band {i+1}: {'Very High' if i==0 else 'High' if i==1 else 'Mid' if i==2 else 'Low' if i==3 else 'Very Low'} Freq",
) for i in range(5)
]
gr.Markdown("### Feature Index Reference:")
for idx, name in enumerate(editor.feature_names):
gr.Markdown(f"- {idx}: {name}")
with gr.Column(scale=1.2):
gr.Markdown("""
### Plot Legend
- **Points with Error Bars**: Observed values where:
- Point position = observed value
- Error bar size = uncertainty (inversely proportional to weight)
- **Green Line**: Model prediction
- **Vertical Red Lines**: Peak points (if enabled)
""")
plots = [gr.Plot() for _ in range(editor.feature_dim)]
# - **Shaded Area**: General prediction uncertainty
def update_scaling_callback(revenue_scale, download_scale, au_scale, show_normalized):
figs, metrics = editor.update_scaling(
revenue_scale,
download_scale,
au_scale,
show_normalized
)
return [*figs, metrics]
def update_model_callback(
data_points_df,
point_groups_df,
enable_area_control,
area_selections,
enable_auc,
auc,
auc_weight,
enable_peaks,
peak_points,
peak_alpha,
peak_weight,
enable_trending,
enable_trending_with_diff,
trending_params
):
figs, _, _, metrics = editor.update_model(
plots,
data_points_df,
point_groups_df,
enable_area_control,
area_selections,
enable_auc,
auc,
enable_peaks,
peak_points,
peak_alpha,
auc_weight,
peak_weight,
enable_trending,
enable_trending_with_diff,
trending_params
)
return [*figs, metrics]
# Update the click handler
update_model_btn.click(
fn=update_model_callback,
inputs=[
data_points_df,
point_groups_df,
enable_area_control,
area_selections,
enable_auc,
auc_input,
auc_weight_input,
enable_peaks,
peak_points_input,
peak_alpha_input,
peak_weight_input,
enable_trending_control,
enable_trending_control_with_diff,
trending_control
],
outputs=[*plots, metrics_display]
)
def update_additional_callback(enable_direct_area, direct_areas):
figs, metrics = editor.update_additional_edit(
enable_direct_area,
direct_areas
)
return [*figs, metrics]
def update_freq_band(band_idx, value):
figs, metrics = editor.update_frequency_bands(band_idx, value)
return [*figs, metrics]
update_scaling_btn.click(
fn=update_scaling_callback,
inputs=[
revenue_scale,
download_scale,
au_scale,
show_normalized
],
outputs=[*plots, metrics_display]
)
update_additional_btn.click(
fn=update_additional_callback,
inputs=[enable_direct_area, direct_areas],
outputs=[*plots, metrics_display]
)
# Add event handlers for frequency band sliders
for i, slider in enumerate(freq_bands):
slider.change(
fn=update_freq_band,
inputs=[gr.Number(value=i, visible=False), slider],
outputs=[*plots, metrics_display]
)
app.load(
fn=update_model_callback,
inputs=[
data_points_df,
point_groups_df,
enable_area_control,
area_selections,
enable_auc,
auc_input,
auc_weight_input,
enable_peaks,
peak_points_input,
peak_alpha_input,
peak_weight_input,
enable_trending_control,
enable_trending_control_with_diff,
trending_control
],
outputs=[*plots, metrics_display]
)
return app
class FunctionParser:
def __init__(self):
# Define available mathematical functions and constants
self.math_functions = {
'sin': np.sin,
'cos': np.cos,
'tan': np.tan,
'exp': np.exp,
'log': np.log,
'sqrt': np.sqrt,
'abs': np.abs,
'pow': np.power,
'pi': np.pi,
'e': np.e,
'asin': np.arcsin,
'acos': np.arccos,
'atan': np.arctan,
'sinh': np.sinh,
'cosh': np.cosh,
'tanh': np.tanh
}
def validate_expression(self, expression: str) -> bool:
"""
Validate the mathematical expression for basic syntax errors.
"""
# Check for balanced parentheses
if expression.count('(') != expression.count(')'):
raise ValueError("Unbalanced parentheses in expression")
# Check for invalid characters
valid_chars = set('0123456789.+-*/()^ xXepi,')
valid_chars.update(''.join(self.math_functions.keys()))
if not all(c in valid_chars or c.isspace() for c in expression.lower()):
raise ValueError("Expression contains invalid characters")
return True
def preprocess_expression(self, expression: str) -> str:
"""
Preprocess the expression to handle various input formats.
"""
# Remove whitespace
expression = expression.replace(' ', '')
# Convert ^ to ** for exponentiation
expression = expression.replace('^', '**')
# Ensure multiplication is explicit
expression = re.sub(r'(\d+)([a-zA-Z])', r'\1*\2', expression)
expression = re.sub(r'(\))([\w])', r'\1*\2', expression)
# Replace X with x for consistency
expression = expression.lower()
return expression
def string_to_function(self, expression: str) -> Callable[[Union[float, np.ndarray]], Union[float, np.ndarray]]:
"""
Convert a string mathematical expression to a callable function.
Args:
expression (str): Mathematical expression (e.g., "sin(x) + x^2")
Returns:
Callable: A function that takes x as input and returns the evaluated result
Example:
>>> f = string_to_function("sin(x) + x^2")
>>> f(0.5)
0.729321...
"""
# Validate and preprocess the expression
self.validate_expression(expression)
processed_expr = self.preprocess_expression(expression)
# Create the function namespace
namespace = self.math_functions.copy()
try:
# Create the lambda function
func = eval(f"lambda x: {processed_expr}", namespace)
# Test the function with a simple input
test_value = 1.0
try:
func(test_value)
except Exception as e:
raise ValueError(f"Invalid function: {str(e)}")
return func
except SyntaxError as e:
raise ValueError(f"Invalid expression syntax: {str(e)}")
except Exception as e:
raise ValueError(f"Error creating function: {str(e)}")
@staticmethod
def demonstrate_usage():
"""
Demonstrate various uses of the function parser.
"""
parser = FunctionParser()
# Test cases
test_expressions = [
"x^2 + 2*x + 1",
"sin(x) + cos(x)",
"exp(-x^2)",
"log(x + 1)",
"sqrt(1 - x^2)",
]
print("Testing various mathematical expressions:")
x_test = 0.5
for expr in test_expressions:
try:
print(f"\nExpression: {expr}")
func = parser.string_to_function(expr)
result = func(x_test)
print(f"f({x_test}) = {result}")
# Test with numpy array
x_array = np.linspace(0, 1, 5)
results = func(x_array)
print(f"f(array) = {results}")
except Exception as e:
print(f"Error: {str(e)}")
# Example usage:
if __name__ == "__main__":
import os
import torch
import numpy as np
# assert torch.cuda.is_available(), "CUDA must be available"
os.environ["WANDB_ENABLED"] = "false"
print(os.getcwd())
device = torch.device(f"cuda:0") if torch.cuda.is_available() else "cpu"
print(f"Device: {device}")
print(f"Using device: {device}")
from models.Tiffusion import tiffusion
model = tiffusion.Tiffusion(
seq_length=365,
feature_size=3,
n_layer_enc=6,
n_layer_dec=4,
d_model=128,
timesteps=500,
sampling_timesteps=200,
loss_type='l1',
beta_schedule='cosine',
n_heads=8,
mlp_hidden_times=4,
attn_pd=0.0,
resid_pd=0.0,
kernel_size=1,
padding_size=0,
control_signal=[]
).to(device)
model.load_state_dict(torch.load("./weight/checkpoint-10.pt", map_location=device, weights_only=True)["model"])
coef = 1.0e-2
stepsize = 5.0e-2
sampling_steps = 100 # Adjustable between 100-500 for speed/accuracy tradeoff
seq_length = 365
feature_dim = 3
print(f"seq_length: {seq_length}, feature_dim: {feature_dim}")
editor = TimeSeriesEditor(seq_length, feature_dim, model)
editor.coef = coef
editor.stepsize = stepsize
editor.sampling_steps = sampling_steps
app = create_gradio_interface(editor)
app.launch(show_api=False)
|