Spaces:
Running
Running
Fix colorbar
Browse files- src/utils.py +80 -19
src/utils.py
CHANGED
@@ -53,12 +53,22 @@ MODEL_CONFIG = {
|
|
53 |
"sundial-base": ("thuml/sundial-base-128m", "Tsinghua University", True, "DL"),
|
54 |
"ttm-r2": ("ibm-granite/granite-timeseries-ttm-r2", "IBM", True, "DL"),
|
55 |
# Task-specific models
|
56 |
-
"stat. ensemble": (
|
|
|
|
|
|
|
|
|
|
|
57 |
"autoarima": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
58 |
"autotheta": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
59 |
"autoets": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
60 |
"seasonalnaive": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
61 |
-
"seasonal naive": (
|
|
|
|
|
|
|
|
|
|
|
62 |
"drift": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
63 |
"naive": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
64 |
}
|
@@ -130,7 +140,10 @@ def format_leaderboard(df: pd.DataFrame):
|
|
130 |
df["zero_shot"] = df["model_name"].apply(get_zero_shot_status)
|
131 |
# Format leakage column: convert to int for all models, 0 for non-zero-shot
|
132 |
df["training_corpus_overlap"] = df.apply(
|
133 |
-
lambda row: int(round(row["training_corpus_overlap"] * 100))
|
|
|
|
|
|
|
134 |
)
|
135 |
df["link"] = df["model_name"].apply(get_model_link)
|
136 |
df["org"] = df["model_name"].apply(get_model_organization)
|
@@ -150,7 +163,12 @@ def format_leaderboard(df: pd.DataFrame):
|
|
150 |
return (
|
151 |
df.style.map(highlight_model_type_color, subset=["model_name"])
|
152 |
.map(lambda x: "font-weight: bold", subset=["zero_shot"])
|
153 |
-
.apply(
|
|
|
|
|
|
|
|
|
|
|
154 |
)
|
155 |
|
156 |
|
@@ -164,12 +182,18 @@ def construct_bar_chart(df: pd.DataFrame, col: str, metric_name: str):
|
|
164 |
alt.Tooltip(f"{col}_upper:Q", title="95% CI Upper", format=".2f"),
|
165 |
]
|
166 |
|
167 |
-
base_encode = {
|
|
|
|
|
|
|
168 |
|
169 |
bars = (
|
170 |
alt.Chart(df)
|
171 |
.mark_bar(color=COLORS["bar_fill"], cornerRadius=4)
|
172 |
-
.encode(
|
|
|
|
|
|
|
173 |
)
|
174 |
|
175 |
error_bars = (
|
@@ -207,7 +231,9 @@ def construct_pairwise_chart(df: pd.DataFrame, col: str, metric_name: str):
|
|
207 |
for c in [col, f"{col}_lower", f"{col}_upper"]:
|
208 |
df[c] *= 100
|
209 |
|
210 |
-
model_order =
|
|
|
|
|
211 |
|
212 |
tooltip = [
|
213 |
alt.Tooltip("model_1:N", title="Model 1"),
|
@@ -218,34 +244,56 @@ def construct_pairwise_chart(df: pd.DataFrame, col: str, metric_name: str):
|
|
218 |
]
|
219 |
|
220 |
base = alt.Chart(df).encode(
|
221 |
-
x=alt.X(
|
|
|
|
|
|
|
|
|
|
|
222 |
y=alt.Y("model_1:N", sort=model_order, title="Model 1"),
|
223 |
)
|
224 |
|
225 |
heatmap = base.mark_rect().encode(
|
226 |
color=alt.Color(
|
227 |
f"{col}:Q",
|
228 |
-
legend=
|
229 |
-
scale=alt.Scale(
|
|
|
|
|
|
|
|
|
|
|
230 |
),
|
231 |
tooltip=tooltip,
|
232 |
)
|
233 |
|
234 |
text_main = base.mark_text(dy=-8, fontSize=8, baseline="top", yOffset=5).encode(
|
235 |
text=alt.Text(f"{col}:Q", format=".1f"),
|
236 |
-
color=alt.condition(
|
|
|
|
|
|
|
|
|
237 |
tooltip=tooltip,
|
238 |
)
|
239 |
|
240 |
return (
|
241 |
(heatmap + text_main)
|
242 |
-
.properties(
|
|
|
|
|
|
|
|
|
|
|
|
|
243 |
.configure_axis(labelFontSize=11, titleFontSize=13, titleFontWeight="bold")
|
244 |
.resolve_scale(color="independent")
|
245 |
)
|
246 |
|
247 |
|
248 |
-
def construct_pivot_table_from_df(
|
|
|
|
|
249 |
"""Construct styled pivot table from precomputed DataFrame."""
|
250 |
|
251 |
def highlight_by_position(styler):
|
@@ -265,7 +313,8 @@ def construct_pivot_table_from_df(errors: pd.DataFrame, metric_name: str) -> pd.
|
|
265 |
|
266 |
if style_parts:
|
267 |
styler = styler.map(
|
268 |
-
lambda x, s="; ".join(style_parts): s,
|
|
|
269 |
)
|
270 |
return styler
|
271 |
|
@@ -273,11 +322,20 @@ def construct_pivot_table_from_df(errors: pd.DataFrame, metric_name: str) -> pd.
|
|
273 |
|
274 |
|
275 |
def construct_pivot_table(
|
276 |
-
summaries: pd.DataFrame,
|
|
|
|
|
|
|
277 |
) -> pd.io.formats.style.Styler:
|
278 |
-
errors = fev.pivot_table(
|
|
|
|
|
279 |
train_overlap = (
|
280 |
-
fev.pivot_table(
|
|
|
|
|
|
|
|
|
281 |
.fillna(False)
|
282 |
.astype(bool)
|
283 |
)
|
@@ -312,12 +370,15 @@ def construct_pivot_table(
|
|
312 |
style_parts.append(f"color: {COLORS['leakage_impute']}")
|
313 |
elif is_imputed_baseline.loc[row_idx, col_idx]:
|
314 |
style_parts.append(f"color: {COLORS['failure_impute']}")
|
315 |
-
elif not style_parts or (
|
|
|
|
|
316 |
style_parts.append(f"color: {COLORS['text_default']}")
|
317 |
|
318 |
if style_parts:
|
319 |
styler = styler.map(
|
320 |
-
lambda x, s="; ".join(style_parts): s,
|
|
|
321 |
)
|
322 |
return styler
|
323 |
|
|
|
53 |
"sundial-base": ("thuml/sundial-base-128m", "Tsinghua University", True, "DL"),
|
54 |
"ttm-r2": ("ibm-granite/granite-timeseries-ttm-r2", "IBM", True, "DL"),
|
55 |
# Task-specific models
|
56 |
+
"stat. ensemble": (
|
57 |
+
"https://nixtlaverse.nixtla.io/statsforecast/",
|
58 |
+
"β",
|
59 |
+
False,
|
60 |
+
"ST",
|
61 |
+
),
|
62 |
"autoarima": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
63 |
"autotheta": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
64 |
"autoets": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
65 |
"seasonalnaive": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
66 |
+
"seasonal naive": (
|
67 |
+
"https://nixtlaverse.nixtla.io/statsforecast/",
|
68 |
+
"β",
|
69 |
+
False,
|
70 |
+
"ST",
|
71 |
+
),
|
72 |
"drift": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
73 |
"naive": ("https://nixtlaverse.nixtla.io/statsforecast/", "β", False, "ST"),
|
74 |
}
|
|
|
140 |
df["zero_shot"] = df["model_name"].apply(get_zero_shot_status)
|
141 |
# Format leakage column: convert to int for all models, 0 for non-zero-shot
|
142 |
df["training_corpus_overlap"] = df.apply(
|
143 |
+
lambda row: int(round(row["training_corpus_overlap"] * 100))
|
144 |
+
if row["zero_shot"] == "β"
|
145 |
+
else 0,
|
146 |
+
axis=1,
|
147 |
)
|
148 |
df["link"] = df["model_name"].apply(get_model_link)
|
149 |
df["org"] = df["model_name"].apply(get_model_organization)
|
|
|
163 |
return (
|
164 |
df.style.map(highlight_model_type_color, subset=["model_name"])
|
165 |
.map(lambda x: "font-weight: bold", subset=["zero_shot"])
|
166 |
+
.apply(
|
167 |
+
lambda x: [
|
168 |
+
"background-color: #f8f9fa" if i % 2 == 1 else "" for i in range(len(x))
|
169 |
+
],
|
170 |
+
axis=0,
|
171 |
+
)
|
172 |
)
|
173 |
|
174 |
|
|
|
182 |
alt.Tooltip(f"{col}_upper:Q", title="95% CI Upper", format=".2f"),
|
183 |
]
|
184 |
|
185 |
+
base_encode = {
|
186 |
+
"y": alt.Y("model_name:N", title="Forecasting Model", sort=None),
|
187 |
+
"tooltip": tooltip,
|
188 |
+
}
|
189 |
|
190 |
bars = (
|
191 |
alt.Chart(df)
|
192 |
.mark_bar(color=COLORS["bar_fill"], cornerRadius=4)
|
193 |
+
.encode(
|
194 |
+
x=alt.X(f"{col}:Q", title=f"{label} (%)", scale=alt.Scale(zero=False)),
|
195 |
+
**base_encode,
|
196 |
+
)
|
197 |
)
|
198 |
|
199 |
error_bars = (
|
|
|
231 |
for c in [col, f"{col}_lower", f"{col}_upper"]:
|
232 |
df[c] *= 100
|
233 |
|
234 |
+
model_order = (
|
235 |
+
df.groupby("model_1")[col].mean().sort_values(ascending=False).index.tolist()
|
236 |
+
)
|
237 |
|
238 |
tooltip = [
|
239 |
alt.Tooltip("model_1:N", title="Model 1"),
|
|
|
244 |
]
|
245 |
|
246 |
base = alt.Chart(df).encode(
|
247 |
+
x=alt.X(
|
248 |
+
"model_2:N",
|
249 |
+
sort=model_order,
|
250 |
+
title="Model 2",
|
251 |
+
axis=alt.Axis(orient="top", labelAngle=-90),
|
252 |
+
),
|
253 |
y=alt.Y("model_1:N", sort=model_order, title="Model 1"),
|
254 |
)
|
255 |
|
256 |
heatmap = base.mark_rect().encode(
|
257 |
color=alt.Color(
|
258 |
f"{col}:Q",
|
259 |
+
legend=None,
|
260 |
+
scale=alt.Scale(
|
261 |
+
scheme=HEATMAP_COLOR_SCHEME,
|
262 |
+
domain=domain,
|
263 |
+
domainMid=domain_mid,
|
264 |
+
clamp=True,
|
265 |
+
),
|
266 |
),
|
267 |
tooltip=tooltip,
|
268 |
)
|
269 |
|
270 |
text_main = base.mark_text(dy=-8, fontSize=8, baseline="top", yOffset=5).encode(
|
271 |
text=alt.Text(f"{col}:Q", format=".1f"),
|
272 |
+
color=alt.condition(
|
273 |
+
text_condition,
|
274 |
+
alt.value(COLORS["text_white"]),
|
275 |
+
alt.value(COLORS["text_black"]),
|
276 |
+
),
|
277 |
tooltip=tooltip,
|
278 |
)
|
279 |
|
280 |
return (
|
281 |
(heatmap + text_main)
|
282 |
+
.properties(
|
283 |
+
height=550,
|
284 |
+
title={
|
285 |
+
"text": f"Pairwise {cbar_label} ({metric_name}) with 95% CIs",
|
286 |
+
"fontSize": 16,
|
287 |
+
},
|
288 |
+
)
|
289 |
.configure_axis(labelFontSize=11, titleFontSize=13, titleFontWeight="bold")
|
290 |
.resolve_scale(color="independent")
|
291 |
)
|
292 |
|
293 |
|
294 |
+
def construct_pivot_table_from_df(
|
295 |
+
errors: pd.DataFrame, metric_name: str
|
296 |
+
) -> pd.io.formats.style.Styler:
|
297 |
"""Construct styled pivot table from precomputed DataFrame."""
|
298 |
|
299 |
def highlight_by_position(styler):
|
|
|
313 |
|
314 |
if style_parts:
|
315 |
styler = styler.map(
|
316 |
+
lambda x, s="; ".join(style_parts): s,
|
317 |
+
subset=pd.IndexSlice[row_idx:row_idx, col_idx:col_idx],
|
318 |
)
|
319 |
return styler
|
320 |
|
|
|
322 |
|
323 |
|
324 |
def construct_pivot_table(
|
325 |
+
summaries: pd.DataFrame,
|
326 |
+
metric_name: str,
|
327 |
+
baseline_model: str,
|
328 |
+
leakage_imputation_model: str,
|
329 |
) -> pd.io.formats.style.Styler:
|
330 |
+
errors = fev.pivot_table(
|
331 |
+
summaries=summaries, metric_column=metric_name, task_columns=["task_name"]
|
332 |
+
)
|
333 |
train_overlap = (
|
334 |
+
fev.pivot_table(
|
335 |
+
summaries=summaries,
|
336 |
+
metric_column="trained_on_this_dataset",
|
337 |
+
task_columns=["task_name"],
|
338 |
+
)
|
339 |
.fillna(False)
|
340 |
.astype(bool)
|
341 |
)
|
|
|
370 |
style_parts.append(f"color: {COLORS['leakage_impute']}")
|
371 |
elif is_imputed_baseline.loc[row_idx, col_idx]:
|
372 |
style_parts.append(f"color: {COLORS['failure_impute']}")
|
373 |
+
elif not style_parts or (
|
374 |
+
len(style_parts) == 1 and "font-weight" in style_parts[0]
|
375 |
+
):
|
376 |
style_parts.append(f"color: {COLORS['text_default']}")
|
377 |
|
378 |
if style_parts:
|
379 |
styler = styler.map(
|
380 |
+
lambda x, s="; ".join(style_parts): s,
|
381 |
+
subset=pd.IndexSlice[row_idx:row_idx, col_idx:col_idx],
|
382 |
)
|
383 |
return styler
|
384 |
|